1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.41
3 * optimization: the default implementation of
4 COMPUTE-DISCRIMINATING-FUNCTION does much less wasted work.
5 * enhancement: Explicit memory barrier operations are now available for use
6 by multithreaded code. See documentation for details.
8 changes in sbcl-1.0.41 relative to sbcl-1.0.40:
9 * optimization: validity of observed keyword initargs to MAKE-INSTANCE is
10 cached, leading to many fewer expensive calls to
11 COMPUTE-APPLICABLE-METHODS.
12 * optimization: in the (unoptimized) general method for MAKE-INSTANCE on a
13 CLASS argument, search for and call an appropriate optimized ctor function
15 * bug fix: WRITE always returns the correct value.
17 changes in sbcl-1.0.40 relative to sbcl-1.0.39:
18 * bug fix: readdir now works on :inode64 darwin builds (lp#592897)
19 * bug fix: Name conflicts between symbols passed as arguments to a single
20 call to IMPORT no longer add multiple symbols with the same name to the
21 package (detectable via DO-SYMBOLS).
22 * bug fix: support building without the dlshim on darwin x86 and x86-64
24 * bug fix: TRACE :ENCAPSULATE NIL now works on ppc/linux.
26 changes in sbcl-1.0.39 relative to sbcl-1.0.38:
27 * bug fix: Backtrace from undefined function on x86 and x86-64 now show
29 * bug fix: linkage-table entries on PPC now no longer overflow their
30 space allocation (potentially causing crashes if they are written out
32 * bug fix: Scrub control stack after scavenging in gencgc on non-x86oid
33 platforms, preventing the GC from seeing stale pointers on the control
34 stack in subsequent GCs (which would, and does, break invariants).
35 * bug fix: 32-bit unicode external formats now work on big-endian systems.
36 * bug fix: Literal characters with code points greater than about 32767
37 now work on PPC UNICODE builds.
38 * bug fix: Any noise left by SSE operations (sqrt and conversions) in the
39 high order bits are explicitly cleared out. In some contrived situations,
40 this could lead to wrong results in mixed real/complex float arithmetic.
41 * bug fix: Fix function/macro redefinition warnings when building with
42 clisp. (lp#576787, thanks to Josh Elsasser)
43 * new platform: experimental support for ppc/openbsd (thanks to Josh
45 * bug fix: Floating-point traps now work on ppc/linux.
47 changes in sbcl-1.0.38 relative to sbcl-1.0.37:
48 * incompatible change: Thread names are now restricted to SIMPLE-STRINGs
49 like for any other thread-related datastructure, MUTEX, etc. (lp#547095)
50 * deprecation: the SB-QUEUE contrib was merged into the SB-CONCURRENCY
51 contrib module. New code should depend on SB-CONCURRENCY, not SB-QUEUE.
52 * deprecation: SB-THEAD:GET-MUTEX was deprecated in favor of
54 * new contributed module: SB-CONCURRENCY is a new contrib; it's supposed to
55 contain additional data structures and tools for concurrent programming;
56 at the moment it contains a lock-free queue, and a lock-free mailbox
58 * new feature: added SB-THREAD:GRAB-MUTEX; it's like the now deprecated
59 GET-MUTEX but takes &key rather than &optional parameters. Also added
60 :TIMEOUT argument to GRAB-MUTEX on non-sb-lutex platforms like Linux and
62 * new feature: added SB-THREAD:TRY-SEMAPHORE, a non-blocking variant of
63 SB-THREAD:WAIT-ON-SEMAPHORE.
64 * new feature: SB-EXT:ATOMIC-DECF has been added as a companion to
66 * new feature: a CANCEL-DEADLINE is associated with DEADLINE-TIMEOUT
67 conditions to defer the deadline for forever.
68 * enhancement: *STANDARD-OUTPUT*, *STANDARD-INPUT*, and *ERROR-OUTPUT* are
70 * enhancement: errors from NO-APPLICABLE-METHOD and
71 NO-PRIMARY-METHOD now have a RETRY restart available to retry the
72 generic function call.
73 * enhancement: SB-BSD-SOCKET improvements
74 ** sockets and socket streams now have a more informative printed
75 representation based on the corresponding SOCKET-NAME and
77 ** SOCKET-MAKE-STREAM once more supports the :AUTO-CLOSE option.
79 ** SOCKET-CLOSE now accepts :ABORT argument, which is passed on to
80 CL:CLOSE when appropriate, and no longer disassociates the stream
81 from the socket if close failed. (lp#543951)
82 * improvements to the instrumenting profiler
83 ** new feature: report per-function GC overhead. (thanks to John Fremlin)
84 ** optimization: counters no longer use locks for the overflow mode.
85 ** bug fix: whenever a profiling counter wrapped into overflow mode, it
86 incurred an off-by-one miscount.
87 * enhancement: improved MAKE-HASH-TABLE documentation (lp#543473)
88 * enhancement: improved DEFMETHOD pretty-printing.
89 * enhancement: perform range reduction when arguments are too large for
90 x87's transcendentals (instead of returning 0). (lp#327192)
91 * enhancement: eliminate some spurious TYPE-WARNINGs. Should help with
92 some of CL-PPCRE's macros. (lp#570079)
93 * enhancement: our machine code is slightly less hostile to valgrind on
95 * enhancement: up-to-date versions of NetBSD-current are supported. (Thanks
96 to Robert Swindells and Aleksej Saushev)
97 * bug fix: correct restart text for the continuable error in MAKE-PACKAGE.
98 * bug fix: a rare case of startup-time page table corruption.
99 * bug fix: a semaphore with multiple waiters and some of them unwinding due
100 to timeouts could be left in an inconsistent state.
101 * bug fix: fix typo in "Reporting Bugs" section of the manual (lp#520366)
102 * bug fix: misoptimization of multiplication by one in
103 (SB-C::FLOAT-ACCURACY 0) policies.
104 * bug fix: miscounts in SB-PROFILE.
105 * bug fix: Fix lost wakeup bug between SB-THREAD:CONDITION-WAIT and
106 CONDITION-NOTIFY on Linux. See threads "lost wakeup in condition-wait /
107 condition-notify" (Feb 2010) and "Condition-Wait, Deadline handler, waking
108 up itself" (March 2010) for further details.
109 * bug fix: allow forward FIND and POSITION on lists to elide checking :END
110 against length of the list if the element is found before the specified
111 END is reached. (thanks to Alec Berryman, lp#554385)
112 * bug fix: errors signalled during package graph modification no longer
113 block FIND-SYMBOL and FIND-PACKAGE in other threads. (lp#511072)
114 * bug fix: SB-POSIX build was broken when SBCL was compiled without the
115 :SB-DOC feature. (lp#552564)
116 * bug fix: SB-INTROSPECT build issues on GENGC/PPC. (lp#490490)
117 * bug fix: more robust runtime executable path detection. (lp#375549)
118 * bug fix: GCD always returns positive values. (lp#413680)
119 * bug fix: Converting division to multiplication by reciprocal handles
121 * bug fix: We were too eager in eliding range reduction tests on x87.
122 The maximal magnitude is 2^63, not 2^64.
123 * bug fix: Transforms for TRUNCATE don't die when the result is completely
125 * bug fix: Maybe restore buildability on Alpha.
126 * bug fix: READ-BYTE isn't inline anymore, fixing weird streams failures.
128 * bug fix: RANDOM-STATE can be printed readably again.
129 * bug fix: Unreadable objects were sometimes printed like #<\nFoo>.
130 * bug fix: Using EQL with non-constant values of constant type (e.g. EQL
131 types) could result in type mismatches during compilation.
133 changes in sbcl-1.0.37 relative to sbcl-1.0.36:
134 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
135 stack frame thrown from.
136 * enhancement: WITH-COMPILATION-UNIT :POLICY allows restricting changes to
137 compiler optimization qualities inside dynamic extent of its body.
138 * enhancement: LOAD-LOGICAL-PATHNAME-TRANSLATIONS can be used to load
139 translations from SYS:SITE;<HOST>.TRANSLATIONS.NEWEST (thanks to Michael
141 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
142 constraint propgation, allowing better compilation eg. when used to
143 access structures with WITH-SLOTS. (lp#520366)
144 * optimization: the compiler is now more aware of the type of the underlying
145 storage vector for multidimensional simple arrays resulting in better code
146 for accessing such arrays.
147 * optimization: passing NIL as the environment argument to TYPEP no longer
148 inhibits optimizing it. (lp#309788)
149 * optimization: more efficient register usage when handling single-float
150 arguments on x86-64. (thanks to Lutz Euler)
151 * optimization: ADJUST-ARRAY and STABLE-SORT on vectors no longer use
152 pre-allocated temporary vectors. (lp#496249)
153 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
154 in :INITIAL-CONTENTS. (lp#523612)
155 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
156 functions. (lp#524707)
157 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
158 WITH-SLOTS usages during compilation.
159 * bug fix: SB-C::CLASS-INFO now prints correctly. (lp#514762)
160 * enhancement: Can now build with ud2 instead of int3 as trap instruction on
161 all x86oid platforms with :UD2-BREAKPOINTS target feature.
162 * bug fix: Breakpoints now work when using ud2 instead of int3 as trap
163 instruction (tested on x86oid linux with ud2-breakpoints).
164 * bug fix: slam.sh now works on win32.
165 * bug fix: better differences of numeric types. (lp#309124)
166 * bug fix: arrays declared intersection and union types can have their
167 upgraded element type derived. (lp#316078)
168 * bug fix: SB-SPROF allocation profiling for all threads failed to profile
169 threads started during profiling. (lp#472499)
170 * bug fix: SB-INTROSPECT test failure when building without SB-EVAL feature.
172 * bug fix: SB-CLTL2:DECLARATION-INFORMATION did not take
173 SB-EXT:RESTRICT-COMPILER-POLICY into account. (lp#313337)
174 * bug fix: Comma inside a backquoted array or structure resulted in nonsense
175 values instead of signaling an error. (lp#309093)
176 * bug fix: Spurious unused variable warning in a DEFSTRUCT edge case.
178 * bug fix: More consistent warnings and notes for ignored DYNAMIC-EXTENT
179 declarations (lp#497321)
180 * bug fix: FIND and POSITION on lists did not check sequence bounds properly
181 and failed to detect circular lists (lp#452008)
182 * bug fix: leakage from ~/.asdf-install into the ASDf-INSTALL contrib build
184 * bug fix: LOOP OF-TYPE VECTOR compile-time error. (lp#540186)
185 * bug fix: SIGNAL SB-SYS:INTERACTIVE-INTERRUPT before entering the debugger
186 due to it, so that handlers can run.
187 * bug fix: reparsing undefined types if they have become defined since
189 * bug fix: missing &REST type in a proclamation for a function with both
190 &REST and &KEY in lambda-list caused miscompilation (lp#458354)
191 * bug fix: WHO-CALLS information for source-transformed and compiler-macro
192 expanded calls (lp#542174)
193 * bug fix: more accurate WHO-MACROEXPANDS information; point into rather
194 than just at toplevel form.
196 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
197 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
198 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
199 but work on type specifiers.
200 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
201 to name a type specifier.
202 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
203 specifier is valid where "valid" basically means "would be accepted as
204 second argument of TYPEP".
205 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
206 returns the function's declared, or derived FTYPE.
207 * new feature: SB-POSIX now supports accessing the d_ino member of
208 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
209 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
210 provide for seeding a RANDOM-STATE object with user-provided data or
211 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
212 attempt to initialize the returned state from the operating system's
213 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
214 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
215 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
216 * bug fix: Fix grammar and style issues for the docstrings of
217 printer-related variables and functions. (Thanks to mon_key; launchpad
219 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
221 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
222 in a row even though a handler defered the deadline long into the
224 * bug fix: A deadline handler was run without interrupts enabled for a
225 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
226 in infinitely spinning, non-killable threads.
227 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
228 before reaching the erring stack frame.
229 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
230 real power. (launchpad bug lp#525949)
231 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
232 GENCGC platforms. (launchpad bug lp#529014)
233 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
234 (launchpad bug lp#525916)
235 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
236 (Thanks to Robert Goldman)
238 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
239 * optimization: ROUND with a single single-float or double-float argument
240 is properly inlined when possible.
241 * optimization: Slightly better code is generated for integer<->float
242 conversions and for single-float<->double-float conversions on x86-64.
243 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
244 code for 32-bit and 64-bit rotations on x86-64.
245 * bug fix: The install script changes the ownership of directories as well
246 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
247 launchpad bug lp#508485)
248 * bug fix: TRUNCATE with a single single-float or double-float argument is
249 properly inlined when possible. (launchpad bug lp#489388)
250 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
251 no longer causes a compiler error on x86 and ppc.
252 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
253 always relative to *READTABLE* rather than the passed argument.
255 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
256 * minor incompatible change: threading support is now enabled by default
258 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
260 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
261 are now caught and reported just like errors during macroexpansion.
262 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
263 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
264 * enhancement: ASDF systems can now depends on SB-QUEUE.
265 * fixes and improvements related to Unicode and external formats:
266 ** bug fix: error handling and restart usage in the ucs-2 external format
268 ** there is now an implementation of the ucs-4 external format.
269 ** the utf-16 and utf-32 external formats are supported.
270 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
271 in the environment. setenv() and unsetenv() are also provided. (reported by
272 Fare Rideau; launchpad bug lp#460455)
273 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
274 loading. This ensures sensible errors for .FASL files from other
275 implementations on case-insensitive filesystems. (reported by Willem
276 Broekema; launchpad bug lp#489417)
277 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
278 Willem Broekema; launchpad bug lp#489698).
279 * bug fix: some minor code rearrangements to reenable warning-free building
280 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
281 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
283 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
286 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
287 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
288 * improvement: support O_LARGEFILE access to files larger than 2GB on
289 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
290 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
291 definitions for methods specializing on the passed class itself.
292 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
293 definitions for methods specializing on the passed class itself, or on
295 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
296 about itself during the build (e.g. for M-? in Slime), if this flag is
297 enabled in customize-target-features.lisp. This will increase the core
298 size by about 5-6mb, though, so it's mostly interesting to SBCL
300 * new feature: various GENCGC tuning parameters have been experimentally
301 documented and exported from SB-EXT. See documentation for details.
302 * fixes and improvements related to Unicode and external formats:
303 ** the Unicode character database has been upgraded to the
304 Unicode 5.2 standard, giving names and properties to a number of new
305 characters, and providing a few extra characters with case
307 ** improvement: restarts for providing replacement input/output on coding
308 errors for fd-stream external formats.
309 ** improvement: where :<encoding> is a keyword corresponding to an
310 external format the system supports, it is now possible to specify
311 (:<encoding> :replacement <character>) as an external format which will
312 automatically substitute <character> on encoding or decoding errors for
313 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
315 ** improvement: the file streams underlying the standard streams (such as
316 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
317 which uses the replacement mechanism to handle encoding errors,
318 preventing various infinite error chains and unrecoverable I/O
320 ** minor incompatible change: the utf-8 external format now correctly
321 refuses to encode Lisp characters in the surrogate range (char-codes
322 between #xd800 and #xdfff).
323 ** fix a typo preventing conversion of strings into octet vectors
324 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
326 ** fix a bug in the octet multibyte handling of decoding errors and the
327 USE-VALUE restart. (launchpad bug lp#314939)
328 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
329 restart on fd-stream decoding errors.
330 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
331 error is near the end of file.
332 ** fix a double-error case in unibyte octet conversions, when the first
333 use of USE-VALUE is ignored.
334 ** fix bugs in handling of undefined code points in unibyte encodings.
335 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
336 streams after an UNREAD-CHAR.
337 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
338 object is allocated in a boxed region of dynamic space.
339 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
340 fork with multiple Lisp threads running is made, instead of going
341 ahead with unpredictable consequences. (reported by Leslie Polzer)
342 * bug fix: uses of slot accessors on specialized method parameters within
343 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
344 error while finalizing the class. This fix may cause classes with slot
345 accessors to be finalized later than previously. (reported by Lars Rune
346 Nøstdal; launchpad bug lp#473699)
347 * bug fix: restore buildability on the MIPS platform. (regression from
348 1.0.30.38, reported by Samium Gromoff)
349 * bug fix: inspecting closures is less likely to fail with a type error.
350 * bug fix: no timer starvation when setting the system clock back.
351 (launchpad bug lp#460283)
352 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
353 standard pprint dispatch table as specified by CLHS.
354 * bug fix: give CLISP a hint about a type declaration to enable it to build
355 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
357 * bug fix: correctly dump literal objects in defaulting forms of arglists.
358 (reported by Attila Lendvai; launchpad bug lp#310132)
359 * bug fix: distinguish in type specifiers between arrays that might be
360 complex and arrays that are definitely complex. (launchpad bug lp#309129)
361 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
362 type. (reported by Levente Mészáros; launchpad bug lp#485972)
363 * bug fix: setting the value of a symbol-macro within a method in the
364 presence of type declarations works properly again. (reported by Iban
365 Hatchondo; launchpad bug lp#485019)
367 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
368 * optimization: faster FIND and POSITION on strings of unknown element type
369 in high SPEED policies. (thanks to Karol Swietlicki)
370 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
372 * improvement: better error signalling for bogus parameter specializer names
373 in DEFMETHOD forms (reported by Pluijzer)
374 * bug fix: DELETE-FILE once again works on logical pathnames (regression
376 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
378 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
379 removes previous default initargs (reported by Lars Rune Nøstdal and
381 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
382 lambda-lists. (reported by Peter Seibel)
383 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
384 correctly when starting from an executable core without saved runtime
385 options (reported by Faré Rideau, thanks to Zach Beane)
386 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
387 an optimization note instead. (reported by Martin Cracauer)
388 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
389 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
390 new process group if it doesn't need to share stdin with the sbcl
391 process. (thanks to Leslie Polzer)
392 * bug fix: SATISFIES could be misoptimized to refer to a local function.
393 (reported by Stanislaw Halik)
395 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
396 * improvement: stack allocation is should now be possible in all nested
397 inlining cases: failure to stack allocate when equivalent code is manually
398 open coded is now considered a bug.
399 * improvements related to Unicode:
400 ** the Unicode character database has been upgraded to the
401 Unicode 5.1 standard, giving names and properties to a number of new
402 characters, and providing a few extra characters with case
404 ** the system now recognizes and produces names for Unicode Hangul
406 ** the EBCDIC-US external-format is now supported for octet operations
407 (as well as for stream operations).
408 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
409 be used to output toplevel compile-time effects into a separate .CFASL
411 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
412 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
414 * optimization: TRUNCATE on known single- and double-floats is upto 25%
416 * optimization: division of floating point numbers by constants uses
417 multiplication by reciprocal when an exact reciprocal exists.
418 * optimization: multiplication of single- and double-floats floats by
419 constant two has been optimized.
420 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
421 sufficient type information is available. (thanks to Leslie Polzer)
422 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
423 known structure objects are as efficient as defstruct generated accessors.
424 * optimization: unused vector creation can now be optimized away.
425 * improvement: ASDF systems can now depends on SB-INTROSPECT.
426 * improvement: a STYLE-WARNING is signalled when a generic function
427 clobbers an earlier FTYPE proclamation.
428 * improvement: the compiler is able to track the effective type of
429 generic function across method addition and removal even in the
430 absence of an explicit DEFGENERIC.
431 * improvement: DESCRIBE now reports on symbols naming undefined
432 but assumed or declared function as well.
433 * improvement: recompilation of systems using SB-GROVEL now works
434 (thanks to Leslie Polzer)
435 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
436 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
438 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
439 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
440 well as user defined declaration names.
441 ** VARIABLE-INFORMATION is now aware of alien variables.
442 * improvement: improved address space layout on OpenBSD (thanks to Josh
444 * improvement: pretty-printing of various Lisp forms has been improved
445 (thanks to Tobias Rittweiler)
446 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
447 unused were deleted in safe code. (reported by John Fremlin)
448 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
450 * bug fix: some out-of-line array predicates were missing (reported by
452 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
454 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
455 (thanks to Larry D'Anna)
456 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
457 as well. (reported by Josh Elsasser)
458 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
459 leading to possible attempts to delete the same file twice. See docstring
460 on DELETE-FILE for details. (reported by John Fremlin)
461 * bug fix: DELETE-FILE once again deletes the file named by the pathname
462 designator argument, rather than its truename. (reported by Luis
464 * bug fix: the low-level debugger had 32-bit assumptions and was missing
465 information about some array types. (thanks to Luis Oliveira)
466 * bug fix: moderately complex combinations of inline expansions could
467 be miscompiled if the result was declared to be dynamic extent.
468 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
469 form (+ <variable> <integer>) were miscompiled under certain
471 * bug fix: in some cases no compiler note about failure to stack allocate
472 was emitted, even if the objects were in fact heap allocated.
473 * bug fix: minor violation of "otherwise inaccessible" rule for stack
474 allocation could cause objects users might reasonably expect to
475 be heap allocated to be stack allocated.
476 * bug fix: DESCRIBE signalled an error for generic functions under
477 certain circumstances. (thanks to Leslie Polzer)
478 * bug fix: Fixed spelling of an error message.
480 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
481 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
482 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
483 of SB-THREAD:THREAD-ERROR-THREAD.
484 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
485 * new feature: docstrings for local and anonymous functions are no longer
486 discarded. (thanks to Leslie Polzer)
487 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
488 values in other threads.
489 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
490 about object allocation.
491 * optimization: division of a real float by a complex float is implemented
492 with a specialised code sequence.
493 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
494 keywords is an order of magnitude faster.
495 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
496 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
497 methods, and similarly for non-standard metaclass classes as long as there
498 are no methods additional on MAKE-INSTANCE.
499 * optimization: more efficient type-checks for FIXNUMs when the value
500 is known to be a signed word on x86 and x86-64.
501 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
502 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
504 * optimization: compiler is smarter about delegating argument type checks to
506 * optimization: several character functions are now compiled somewhat more
507 efficiently. (reported by Lynn Quam)
508 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
509 * improvement: less unsafe constant folding in floating point arithmetic,
510 especially for mixed complex/real -float operations.
511 * optimization: constant double and single floats are stored in native
512 unboxed format on x86[-64].
513 * optimization: smarter code for arithmetic operations with constant floats,
514 complex floats, or integers on x86[-64].
515 * optimization: smarter code for conjugate/multiplication of float complexes
516 and abs/negate of floats on x86-64.
517 * optimization: more efficient complex float and real float operations on
519 * improvement: complex float division is slightly more stable.
520 * improvement: DESCRIBE output has been reworked to be easier to read and
521 contains more pertinent information.
522 * improvement: failure to provide requested stack allocation compiler notes
523 provided in all cases (requested stack allocation not happening without a
524 note being issued is now considered a bug.)
525 * bug fix: SB-POSIX exports the documented types and functions
526 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
527 types. (reported by "abhi")
528 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
529 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
530 (thanks to Paul Khuong)
531 * bug fix: looping around HANDLER-CASE could silently consume stack space
532 on each iteration. (reported by "foobar")
533 * bug fix: better error signalling when calls to functions seeking elements
534 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
535 (reported by Tobias Rittweiler)
536 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
537 directory components now works as it used to. (various prolems reported by
538 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
539 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
540 (reported by Elliot Slaughter, thanks to Stas Boukarev)
541 * bug fix: bogus undefined variable warnings from fopcompiled references to
542 global variables. (thanks to Lars Rune Nøstdal)
543 * bug fix: foreign function names should now appear in backtraces on
544 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
545 * bug fix: SETF compiler macro documentation strings are not discarded
547 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
548 (thanks to Alex Plotnick)
549 * bug fix: SLEEP accepts large integer arguments, truncating them to
550 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
551 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
552 * bug fix: compiler notes for expensive slot type checks could be emitted
553 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
555 changes in sbcl-1.0.29 relative to 1.0.28:
556 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
557 https://bugs.launchpad.net/sbcl
558 Bugs can be reported directly there, or by sending email to
559 sbcl-bugs@lists.sourceforge.net
560 (no subscription required.)
561 * minor incompatible change: under weak type checking policy integer
562 types are weakened less aggressively.
563 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
564 allowed to return, which causes SBCL to quit with exit status 0. Previously
565 if the function returned with a small integer return value, that value
566 was accidentally reused as the exit status.
567 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
568 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
569 :HASH-FUNCTION argument. Refer to user manual for details.
570 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
572 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
573 microseconds since the Unix epoch on all platforms.
574 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
575 allows the compiler to safely elide boundedness checks for special
577 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
578 the symbol, prohibits both lexical and dynamic binding. This is mainly an
579 efficiency measure for threaded platforms, but also valueable in
581 * new feature: UNC pathnames are now understood by the system on Windows.
582 * optimization: the compiler uses a specialized version of FILL when the
583 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
585 * optimization: accesses to potentially non-simple arrays where element type
586 is known are 50% faster.
587 * optimization: compiler now generates faster array typechecking code.
588 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
590 * optimization: multidimensional array accesses in the absence of type
591 information regarding array rank are approximately 10% faster due to
592 open coding of ARRAY-RANK.
593 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
594 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
596 * optimization: result of call to VECTOR can now be stack allocated.
597 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
598 as long as the resulting array is one-dimensional and has a known
599 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
600 length of the list matches the known length of the vector does not
601 allocate the list as an intermediate step. Ditto for VECTOR and simple
603 * optimization: MAKE-ARRAY can now stack allocate in the presence of
604 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
605 known element type, and is known to be simple and one dimensional.
606 * improvement: SBCL now emits a compiler note where stack allocation was
607 requested but could not be provided (not in all cases, unfortunately)
608 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
609 * improvement: pretty-printing loop has been implemented properly. (thanks
610 to Tobias Rittweiler)
611 * documentation: CLOS slot typechecing policy has been documented.
612 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
613 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
614 pointer, leading to memory faults. (thanks to Bart Botta)
615 * bug fix: better floating point exception handling on x86/OpenBSD.
616 (thanks to Josh Elsasser)
617 * bug fix: exit status from QUIT when called under --script was lost
618 (reported by Hubert Kauker)
619 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
620 same implementation of FILL to initialize the array, even if a more
621 efficient one was available (reported by Stas Boukarev, thanks to
623 * bug fix: potential miscompilation of array stack allocation on x86 and
624 x86-64. (reported by Time Tossavainen)
625 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
626 result in their subforms being treated as top level forms. (reported by
628 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
629 than a machine register could leave garbage in the high bits of the
630 result register (bug 316325).
631 * bug fix: disable address space randomization Linux/x86-64 as well,
632 not just x86-64. (reported by Ken Olum)
633 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
634 generate incorrect code.
635 * bug fix: #201; type inference for CONS and ARRAY types could derive
636 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
637 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
638 could not be assigned to or bound with PROGV. (reported by Lorenz
640 * bug fix: the value of CL:- in the inspector was the previous expression
641 evaluated rather than the expression being evaluated.
642 * bug fix: constants can no longer be locally declared special.
643 * bug fix: signals delivered to threads started from foreign land (read:
644 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
645 thread by blocking all signals and resignalling.
646 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
647 from :INITFORM, if any.
649 changes in sbcl-1.0.28 relative to 1.0.27:
650 * a number of bugs in cross-compilation have been fixed, with the ultimate
651 result that building under (at least) clisp should be much more reliable.
652 * minor incompatible changes: echo-streams now propagate unread-char to the
653 underlying input stream, and no longer permit unreading more than one
655 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
656 frames, allowing for instance more reliable backtraces.
657 * improvement: the debugger REPL can now reference lexical variables
658 by name directly for code compiled with (DEBUG 3).
659 * improvement: errors from malformed declarations now have better source
660 paths associated with them. (thanks to Tobias Rittweiler)
661 * optimization: faster local calls on x86/x86-64
662 * bug fix: some error messages for out-of-bound array indexes confused the
663 index and the bound. (thanks to Stas Boukarev)
664 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
666 * bug fix: running regressions tests in shells without OSTYPE set now works.
667 (reported by Harald Hanche-Olsen)
668 * bug fix: more robust static space exhaustion signalling from
669 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
670 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
671 docstring away instead of storing it under names such as (LAMBDA (X)).
672 (reported by Leslie Polzer)
673 * bug fix: timers could go off in the wrong order, be delayed indefinitely
674 (thanks to Ole Arndt for the patch)
675 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
676 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
677 potentially-invalid effective methods in its cache.
678 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
679 instances as well (thanks to Paul Khuong)
680 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
681 SB-POSIX:WAITPID and their C equivalents.
682 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
684 changes in sbcl-1.0.27 relative to 1.0.26:
685 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
686 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
687 * improvement: the system either recovers from stack exhaustion or dies
688 properly as opposed to leaving the user uncertain of whether the handler
689 trampled on some random memory next to the stack or having to rely on
690 --lose-on-corruption (which is still a good idea to use in production
691 because stack exhaustion can happen in signal handlers which will likely
693 * bug fix: fix gc related interrupt handling bug on ppc (regression from
694 1.0.25.37, reported by Harald Hanche-Olsen)
695 * bug fix: work around signal delivery bug in darwin (regression from
696 1.0.25.44, reported by Sidney Markowitz)
697 * bug fix: fix ERROR leaking memory (reported by David Thompson)
699 changes in sbcl-1.0.26 relative to 1.0.25:
700 * incompatible change: an interruption (be it a function passed to
701 INTERRUPT-THREAD or a timer function) runs in an environment where
702 interrupts can be enabled. The interruption can use
703 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
704 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
705 potentially running out of stack. Keep in mind that in the absence
706 of WITHOUT-INTERRUPTS some potentially blocking operation such as
707 acquiring a lock can enable interrupts.
708 * incompatible change: GC-OFF and GC-ON are removed, as they were
709 always unsafe. Use WITHOUT-GCING instead.
710 * new feature: runtime option --disable-ldb
711 * new feature: runtime option --lose-on-corruption to die at the
712 slightest hint of possibly non-recoverable errors: running out of
713 memory, stack, alien stack, binding stack, encountering a memory
714 fault, etc. In the absence of --lose-on-corruption a warning is
716 * enhancement: detect binding stack exhaustion
717 * enhancement: detect alien stack exhaustion on x86/x86-64
718 * improvement: generally more stable and reliable interrupt handling
719 * improvement: there is a per thread interruption queue,
720 interruptions are executed in order of arrival
721 * improvement: a repeating timer reschedules itself when the it has
722 finished, but expiration times are spaced equally. If an
723 expiration time is in the past it will trigger after a short grace
724 period that may give a chance to other things to run.
725 * optimization: slightly faster gc on multithreaded builds
726 * optimization: faster WITHOUT-GCING
727 * bug fix: when JOIN-THREAD signals an error, do it when not holding
728 important locks so that the debugger/handler doesn't produce
729 recursive errors or deadlock.
730 * bug fix: real-time signals are not used anymore, so no more
731 hanging when the system wide real-time signal queue gets full.
732 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
734 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
735 * bug fix: fix random memory faults related to interrupts on alpha
736 * bug fix: fix random memory faults related to interrupts on sparc
737 * bug fix: fix deadlocks related to starting threads
738 * bug fix: fix deadlines on locks on futex platforms
739 * bug fix: restore errno in signal handlers
740 * bug fix: fix deadlocks related to hash tables
741 * bug fix: fix deadlocks in pcl
743 changes in sbcl-1.0.25 relative to 1.0.24:
744 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
745 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
746 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
747 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
748 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
749 better name and does not return values so stale on multiprocessor systems.
750 Also, HOLDING-MUTEX-P was added for about the only sane usage of
752 * improvement: unithread builds keep track of MUTEX-VALUE.
753 * improvement: reading from a TWO-WAY-STREAM does not touch the output
754 stream anymore making it thread safe to have a concurrent reader and
755 a writer, for instance, in a pipe.
756 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
757 arguments, making compiler-macros for SETF-functions able to inspect
758 their constant arguments.
759 * improvement: COMPILE-FILE reports times with millisecond accuracy
760 (thanks to Luis Oliveira)
761 * optimization: CHAR-CODE type derivation has been improved, making
762 TYPEP elimination on subtypes of CHARACTER work better. (reported
763 by Tobias Rittweiler, patch by Paul Khuong)
764 * bug fix: setting alien structure fields of type struct by value now
765 computes the right offset for the memory copy.
766 * bug fix: compilation problem involving inlined calls to aliens with
767 result type VOID. (reported by Ken Olum)
768 * bug fix: #235a; sequential inline expasion in different policies no
769 longer reuses the functional from the previous expansion site.
770 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
771 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
772 * bug fix: named ENUMs in multiply-referenced alien record types no longer
773 cause an implied type redefinition. Regression from 1.0.21.29.
774 * bug fix: sign-extension of small signed return values now works with C
775 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
776 * improvements to the Windows port:
777 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
778 Westfold, thanks to Rudi Schlatte)
780 changes in sbcl-1.0.24 relative to 1.0.23:
781 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
782 vector of a multidimensional SIMPLE-ARRAY.
783 * new feature: the system now signals a continuable error if standard
784 readtable modification is attempted. (thanks to Tobias Rittweiler)
785 * new feature: DIRECTORY has been extended with a non-standard keyword
786 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
787 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
788 has been robustified and documented better. (thanks to Robert Goldman)
789 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
790 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
791 * tradeoff: constant FORMAT control strings are now compiled unless
792 SPACE > SPEED (previously only when SPEED > SPACE.)
793 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
795 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
797 * bug fix: lexical type declarations are now correctly reported by
798 SB-CLTL2. (reported by Larry D'Anna)
799 * bug fix: STRING-TO-OCTETS did not handle :START properly when
800 using UTF-8 as external format. (thanks to Luis Oliveira)
801 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
802 are signalled correctly. (thanks to Stas Boukarev)
803 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
804 designator. (thanks to Tobias Rittweiler)
805 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
806 designator, and returns T instead of the function. (thanks to
808 * bug fix: direct superclasses of STANDARD-CLASS and
809 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
810 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
811 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
812 when no class with the named slot yet exists no longer causes a
813 compile-time style-warning.
814 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
815 in safe code. (reported by Didier Verna)
816 * bug fix: #430; nested structure constructors can stack allocate.
817 * bug fix: on some 64-bit platforms dynamic space size was truncated
818 to #xffff0000 bytes. (reported by Benjamin Lambert)
819 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
820 REPL unusable. (reported by Daniel Herring)
821 * bug fix: return values of READ-SEQUENCE did not take :START into
822 account on file streams, regressions since 1.0.12.22. (reported by
823 Thomas Russ, patch by Paul Khuong)
824 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
825 method specializer used to confuse permuation vector optimization.
826 * bug fix: system inserted bogus implicit type declarations for local
827 special variables in DEFMETHOD bodies.
828 * bug fix: #354; duplicated frames in backtraces due to
829 non-tail-call-optimized XEPs to functions with return type NIL
831 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
832 initializes structure object slots according to DEFSTRUCT initforms,
833 and DEFSTRUCT forms :INCLUDEind structure classes defined using
834 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
835 (reported by Bruno Haible and Stephen Wilson)
836 * bug fix: #395; fill-pointer output streams used now support
837 element-type BASE-CHAR as well.
838 * bug fix: compiler error when attempting to derive return value of
839 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
841 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
843 * bug fix: Mac OS X binaries should now be portable between Leopard
846 changes in sbcl-1.0.23 relative to 1.0.22:
847 * enhancement: when disassembling method functions, disassembly
848 for the associated fast function is also produced.
849 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
851 * optimization: printing with *PRINT-PRETTY* true is now more
852 efficient as long as the object being printed doesn't require
853 special handling by the pretty printer.
854 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
855 now interact correctly with type declarations.
856 * partial bug fix: PCL detects infinite recursion during wrapper
857 validation. (thanks to Attila Lendvai)
858 * bug fix: #426; nested function calls are inlined properly.
859 Previously if FOO was an inline function, in calls of the form
860 (FOO (FOO ...)) the outer call was not inlined.
861 * bug fix: long long arguments passed in registers to alien
862 functions on PowerPC Linux and NetBSD are now handled correctly
863 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
865 changes in sbcl-1.0.22 relative to 1.0.21:
866 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
867 for the shared object in the current directory, but passes the native
868 namestring of the designated pathname to the operation system's shared
869 object loading function as-it.
870 * minor incompatible change: --disable-debugger toplevel option now takes
871 effect before processing of initialization files and --eval or --load
873 * new feature: new commandline argument: --script, which supports
874 shebang lines. See documentation for details. (based on work by
876 * new feature: SAVE-LISP-AND-DIE can save current values of
877 --dynamic-space-size and --control-stack-size in the executable core,
878 causing it to skip normal runtime option processing. See documentation
879 for details. (thanks to Zach Beane)
880 * enhancement: inoccous calls to EVAL or generic functions dispatching
881 on subclasses of eg. STREAM no longer cause compiler notes to appear.
882 * enhancement: the system no longer resignals errors from --load and
883 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
884 associated with the original error to be lost. (thanks to Ariel
886 * enhancement: :DONT-SAVE keyword argument has been added to
887 LOAD-SHARED-OBJECT for controlling interaction with
888 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
889 the effects of an earlier LOAD-SHARED-OBJECT call.
890 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
891 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
892 provided. (reported by Cedric St-Jean)
893 * bug fix: circularity handling in the reader did not treat raw
894 structure slots correctly. (reported by Cedric St-Jean)
895 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
896 file descriptors when there were none.
897 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
898 was broken. (reporteed by Willem Broekema)
899 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
900 pathnames without a directory.
901 * bug fix: COMPILE-FILE did not warn about undefined variable
902 references at toplevel, and LOAD of the resulting fasl did
904 * bug fix: functions with non-required arguments used to end up with
905 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
906 * bug fix: redefining a function with non-required arguments didn't
907 update the system's knowledge about its call signature properly.
908 * bug fix: fixed #431; incompatible alien record type redefinitions
909 are detected and handled. (reported by Neil Haven)
910 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
911 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
913 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
914 after alien stack frames.
915 * bug fix: strings passed to mk{s,d}temp lacked null termination
917 changes in sbcl-1.0.21 relative to 1.0.20:
918 * new feature: the compiler is able to track the effective type of a
919 generic function across method addition and removal.
920 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
921 appropriately typed structure slots without locking.
922 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
923 information like those gathered by TIME using a programming-friendly
925 * new feature: TIME reports time taken even if the form performs a
926 non-local transfer of control.
927 * enhancement: reduced conservativism on GENCGC platforms: on
928 average 45% less pages pinned (measured from SBCL self build).
929 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
930 mutate constant symbols or violate declaimed type of the symbol.
931 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
932 sleeping on the semaphore in heavily contested semaphores.
933 * bug fix: semaphores and condition variables were not interrupt
935 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
936 owned by other threads anymore.
937 * bug fix: FIND on lists called KEY outside the specified
938 subsequence. (reported by budden)
939 * bug fix: LOG doesn't use single-float intermediate results when
940 given mixed integer and double-float arguments, leading to better
941 precision. (reported by Bob Felts)
942 * bug fix: LOG with base zero returned values of inconsistent type.
944 changes in sbcl-1.0.20 relative to 1.0.19:
945 * minor incompatible change: OPTIMIZE qualities
946 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
947 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
948 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
949 * documentation: some slot access efficiency guidelines have been
950 added to the user manual.
951 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
952 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
953 as ASSOC and MEMEBER.
954 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
955 to more efficient EQ-comparison versions more often.
956 * optimization: enhanced derivation of DOLIST iteration variable type
958 * optimization: constant folding of simple (LIST ...) forms as DOLIST
960 * optimization: runtime lookup of function definitions can be
961 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
962 * optimization: compiler is able to derive the return type of
963 (AREF (THE STRING X) Y) as being CHARACTER.
964 * optimization: CLRHASH on empty hash-tables no longer does pointless
965 work. (thanks to Alec Berryman)
966 * optimization: REPLACE deftransforms don't punt when :START1 or
968 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
969 no longer emits a code deletion note. (reported by Andrew
971 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
972 functions with an odd number of &OPTIONAL arguments, a &REST
973 argument, and one or more &KEY arguments at the call site.
974 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
975 arguments appeared at call sites as well.
976 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
977 READ-LINE &co can no longer read from them afterwards. (reported
979 * bug fix: fixed #427: unused local aliens no longer cause compiler
980 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
982 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
983 alien-stack leakage. (reported by Andy Hefner)
984 * bug fix: PROGV signals an error when an attempt to violate declared
985 type of a variable or bind a constant is made.
986 * bug fix: SET signals an error when an attempt to violate declared
987 type of a variable is made.
988 * bug fix: restart computation during the execution of a restart
989 test function no longer causes infinite recursion. (reported by
991 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
992 signals an error. (thanks to Pierre Mai)
993 * bug fix: EXPORT left symbol unexported in conflict situations.
994 (thanks to Michael Weber)
995 * bug fix: correctly handle name conflicts involving CL:NIL.
996 (thanks to Michael Weber)
997 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
998 conflicts arising from USEing package with conflicting symbols
999 correctly. (thanks to Michael Weber)
1001 changes in sbcl-1.0.19 relative to 1.0.18:
1002 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
1003 warnings that go otherwise unhandled will be muffled if they are
1004 of the type that's the value of this variable.
1005 * optimization: stack allocation is slightly more efficient on x86
1007 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
1008 where a raw slot always is initialized using the initform whose
1009 type is not know sufficiently well a compile-time are now compiled
1010 correctly. (reported by John Morrison)
1011 * bug fix: compiler no longer makes erronous assumptions in the
1012 presence of non-foldable SATISFIES types.
1013 * bug fix: stack analysis missed cleanups of dynamic-extent
1014 arguments in non-let-converted calls to local functions.
1015 * improvements to the Windows port:
1016 ** adjusted address spaces for building on both Win32 and
1017 Win64. (thanks for John Connors)
1018 * fixed some bugs revealed by Paul Dietz' test suite:
1019 ** interval arithmetic during type derivation used inexact integer
1020 to single-float coercions.
1021 ** arithmetic operations involving large integers and single
1022 floats give the same results in compiled and interpreted code.
1023 ** deriving the result type of COERCE no longer signals an error
1024 if the derived type of the second argument is a MEMBER type
1025 containing invalid type specifiers.
1026 ** ADJOIN with constant NIL as second argument works correctly.
1027 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
1029 changes in sbcl-1.0.18 relative to 1.0.17:
1030 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
1031 profiles only the current thread.
1032 * minor incompatible change: changes to SYMBOL-VALUE of constants
1033 defined with DEFCONSTANT now signal an error.
1034 * enhancement: SB-SPROF now has support for wallclock profiling,
1035 and is also able to profile specific threads. REPORT output
1036 has also additional sorting options.
1037 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
1039 * optimization: structure allocation has been improved
1040 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
1041 ** out of line constructors are ~10% faster.
1042 ** inline constructors are ~15% faster.
1043 ** inline constructors are capable of dynamic extent allocation
1044 (generally on x86 and x86-64, in some cases on other platforms
1046 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
1048 * optimization: file compiler is now able to coalesce non-circular
1049 lists, non-base strings, and bit-vectors. Additionally, constants
1050 are never referenced through SYMBOL-VALUE at runtime.
1051 * optimization: code defining methods on PRINT-OBJECT (and other
1052 generic functions in the COMMON-LISP package) now loads faster.
1053 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
1055 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
1056 fasl is now deleted. (reported by Attila Lendvai)
1057 * bug fix: READ-LINE always returned NIL for the last line in files.
1058 (reported by Yoshinori Tahara)
1059 * bug fix: more accurate disassembly annotations of foreign function
1060 calls. (thanks to Andy Hefner)
1061 * bug fix: trimming non-simple strings and non-string string
1062 designators when the there is nothing to trim works properly.
1063 (thanks to James Knight)
1064 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
1065 (thanks to Travis Cross)
1066 * fixed some bugs revealed by Paul Dietz' test suite:
1067 ** NIL is a valid function name (regression at 1.0.13.38)
1068 ** FILL on lists was missing its return value (regression at 1.0.12.27)
1069 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1070 fill pointers properly (regression at 1.0.12.23)
1071 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1072 displacement indices properly (regression at 1.0.12.23)
1074 changes in sbcl-1.0.17 relative to 1.0.16:
1075 * temporary regression: user code can no longer allocate closure
1076 variable storage on stack, due to bug 419 without explicitly
1077 requesting it. Please consult sbcl-devel for advice if you need to
1078 use this feature in the meanwhile.
1079 * new feature: runtime argument --control-stack-size can be used to
1080 adjust thread default control stack size.
1081 * enhancement: improved TIME output
1082 ** all times are reported using the measured accuracy (milliseconds
1083 for real and GC times, microseconds for everything else.)
1084 ** processor cycle counts on x86 and x86-64.
1085 ** interpreted forms are counted for both evaluator modes.
1086 ** number of lambdas converted by the compiler is reported.
1087 ** CPU percentage report (computed from real and total run time.)
1088 ** more comprehensive run time reporting, using a condenced format
1089 ** interperted form, lambda, and page fault counts are omitted
1091 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
1093 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
1094 * optimization: two argument forms of LAST are upto ~10% faster
1095 in normal SPEED policies.
1096 * optimization: NCONC no longer needs to heap cons its &REST list
1097 in normal SPEED policies.
1098 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
1099 to Sidney Markowitz)
1100 * bug fix: UNION and NUNION work with :TEST-NOT once more,
1101 regression since 1.0.9.1. (thanks to Eric Marsden)
1102 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
1103 since 1.0.15.36. (thanks to Paul Khuong)
1104 * bug fix: LAST when always returned the whole list when given a bignum
1105 as the second argument.
1106 * bug fix: dynamic extent allocation of nested lists and vectors
1107 could leak to otherwise accessible parts.
1108 * bug fix: invalid optimization of heap-allocated alien variable
1110 * bug fix: fasl header checking is less vulnerable to different
1111 platform word lengths.
1112 * bug fix: more correct assembler syntax for GNU binutils
1113 2.18.50.0.4 support. (thanks to Marijn Schouten)
1114 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
1116 * bug fix: the fopcompiler can handle LOCALLY forms (with no
1117 declarations) successfully. (reported by Attila Lendvai)
1119 changes in sbcl-1.0.16 relative to 1.0.15:
1120 * minor incompatible change: revert the changes to sb-posix's error
1121 signaling added in 1.0.14.
1122 * minor incompatible change: change PROBE-FILE back to returning
1123 NIL whenever we can't get a truename, as was the case before 1.0.14.
1124 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
1125 inherits from ERROR instead of just CONDITION.
1126 * new feature: SB-INTROSPECT can provide source locations for instances
1127 as well. (thanks to Tobias Rittweiler)
1128 * optimization: binding special variables now generates smaller code
1129 on threaded platforms.
1130 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
1131 and cases where no :TEST is given but the compiler can infer that
1132 the element to search is of type (OR FIXNUM (NOT NUMBER)).
1133 * optimization: better LOGNOT on fixnums.
1134 * optimization: modular arithmetic for a particular requested width
1135 is implemented using a tagged representation unless a better
1136 representation is available.
1137 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
1138 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
1139 for wellformedness and returns a specialized vector. (reported by
1140 Francois-Rene Rideau)
1141 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
1142 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
1143 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
1144 unscheduling timer at the same time as another timer fires.
1145 * bug fix: don't reschedule timers for dead threads.
1146 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
1147 * bug fix: copying output from RUN-PROGRAM to a stream signalled
1148 bogus errors if select() was interrupted.
1149 * enhancement: add support for fcntl's struct flock to SB-POSIX.
1151 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
1152 * enhancement: cleaner backtraces for interactive interrupts, as
1153 well as other cases where the interesting frames used to be
1154 obscured by interrupt handling frames.
1155 * enhancement: untracing a whole package using (UNTRACE "FOO") is
1156 now supported, and tracing a whole package using (TRACE "FOO") now
1157 traces SETF-functions as well.
1158 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
1159 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
1160 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
1161 * unreadably printed representation of hash-tables now includes
1163 * bug fix: partially fixed #188: type propagation from assignments
1164 is now more efficient.
1165 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
1166 'DOUBLE-FLOAT) are not flushable.
1167 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
1168 full address of the object, and none of the tag bits.
1169 * bug fix: readably printing hash-tables now respects other printer
1170 control variables. (reported by Cedric St-Jean)
1171 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
1172 keyword with MAKE-HASH-TABLE.
1173 * bug fix: export SB-POSIX:MKSTEMP.
1174 * bug fix: SORT was not interrupt safe.
1175 * bug fix: XREF accounts for the last node of each basic-block as
1177 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1178 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1179 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1180 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1181 create obsolete instances in the case of redefinition or
1182 obsoletion of a superclass. (thanks to Andy Hefner)
1183 * bug fix: Support for the Alpha architecture has been revived; it had
1184 suffered somewhat from lack of maintenance since sbcl-1.0.
1185 * improvements to the Windows port:
1186 ** The system detects the codepage to use at startup, instead of
1187 using the value from the saved core. (thanks to Kei Suzuki)
1189 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1190 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1191 (see documentation for details.)
1192 * revived support for OpenBSD (contributed by Josh Elsasser)
1193 * partially fixed bug #108: ROOM no longer suffers from occasional
1194 (AVER (SAP= CURRENT END)) failures.
1195 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1196 bodies no longer cause a WARNING to be signalled. (reported by
1198 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1199 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1200 the bug in TRUNCATE.) (reported by Michael Weber)
1201 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1202 no samples. (reported by Andy Hefner)
1203 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1204 no longer appear as (NIL ...) frames in backtraces.
1205 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1206 is now actually exported from SB-EXT as documented. (reported by
1208 * bug fix: sb-aclrepl now correctly understands how to inspect
1209 single-floats on 64-bit platforms where single-floats are not boxed.
1210 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1211 yet been finalized. (reported by Levente Meszaros)
1212 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1213 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1214 SBCL-specific optimize qualities.
1216 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1217 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1218 an executable in the search path, and does so in the child
1219 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1220 been removed; it can be found in the CVS history, for somebody who
1221 needs that search behavior (see the manual).
1222 * minor incompatible change: compiler policy re. weakening type
1223 checks has changed: now type checks are weakened only if SAFETY < 2
1225 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1226 unparsing of directory pathnames as files. Analogously,
1227 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1228 filename to parse into a directory pathname.
1229 * enhancement: implicit generic function creation now signals a
1230 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1231 users can bind handlers for and muffle around calls to LOAD.
1232 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1233 to be used (using the default stream external format), and allows
1234 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1235 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1236 strings and vectors whose element-type or simplicity is not fully
1237 known at compile-time.
1238 * optimization: STRING-TRIM and related functions no longer allocate
1239 a new string when no trimming needs to be performed. These functions
1240 are also faster than before when the input string has been declared
1242 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1243 * optimization: READ-LINE is significantly faster for files containing
1245 * optimization: non-open coded uses of character comparison operators
1246 (e.g. char=) no longer cons when called with more than one parameter
1247 on platforms supporting dynamic-extent allocation.
1248 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1249 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1250 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1252 * bug fix: some sequence functions elided bounds checking when
1254 * bug fix: too liberal weakening of union-type checks when SPEED >
1256 * bug fix: more bogus fixnum declarations in ROOM implementation
1259 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1260 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1261 :SYNCHRONIZED argument, which makes the hash-table safe for
1262 concurrent accesses (but not iteration.) See also:
1263 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1264 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1265 * optimization: CONCATENATE on strings is an order of magnitude faster
1266 in code compiled with (> SPEED SPACE).
1267 * optimization: SUBSEQ is ~50% faster on lists.
1268 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1269 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1270 END is smaller then START.
1271 * bug fix: SB-PROFILE will no longer report extra consing for nested
1272 calls to profiled functions.
1273 * bug fix: ROOM implementation had bogus fixnum declarations which
1274 could cause type-errors when calling ROOM on large images.
1275 * bug fix: if file compilation is aborted, the partial fasl is now
1276 deleted, and COMPILE-FILE returns NIL as the primary value.
1277 * bug fix: number of thread safety issues relating to SBCL's internal
1278 hash-table usage have been fixed.
1279 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1280 be returned from its body when the values were being returned
1281 using unknown-values return convection and the W-P-O was wrapped
1282 inside an UNWIND-PROTECT.
1283 * bug fix: sb-posix should now compile again under Windows, enabling
1284 slime to work again.
1286 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1287 * incompatible change: hash-table accessor functions are no longer
1288 automatically protected by locks. Concurrent accesses on the same hash-table
1289 from multiple threads can give inconsistent results or even corrupt the
1290 hash-table completely. Multi-threaded applications should do their own
1291 locking at the correct granularity. In the current implementation it is
1292 still safe to have multiple readers access the same table, but it's not
1293 guaranteed that this property will be maintained in future releases.
1294 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1295 and will signal an error at runtime.
1296 * enhancement: SB-THREAD package now exports a semaphore interface.
1297 * enhancement: CONS can now stack-allocate on x86 and
1298 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1300 * enhancement: nested lists can now be stack allocated on
1301 platforms providing stack allocation support.
1302 * enhancement: dynamic-extent support has been extended to support
1303 cases where there are multiple possible sources for the stack
1305 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1306 if the mutex is uncontested on Linux.
1307 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1308 as specified by the CLHS. (thanks to Tobias Rittweiler)
1309 * bug fix: NaN comparison now works on x86-64.
1310 * bug fix: CLOSE :ABORT T on a stream with pending output now
1312 * bug fix: instances of non-standard metaclasses using standard
1313 instance structure protocol sometimes missed the slot type checks
1315 * bug fix: known functions can be inlined in high-debug code.
1317 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1318 * minor incompatible change: the MSI installer on Windows no longer
1319 associates .lisp and .fasl files with the installed SBCL.
1320 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1321 on Windows. (thanks to Luis Oliviera)
1322 * new feature: SB-CLTL2 contrib module now implements
1323 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1324 * optimization: scavenging weak pointers is now more efficient,
1325 requiring O(1) instead of O(N) per weak pointer to identify
1326 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1327 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1328 method bodies using constant slot-names when the first argument is
1329 a specializer parameter for the method.
1330 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1331 names, when the class of the instance is a direct instance
1332 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1333 as slow as the constant slot-name case.)
1334 * optimization: member type construction is now O(N) instead
1336 * optimization: UNION and NUNION are now O(N+M) for large
1337 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1339 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1340 now more readable in environments like Slime which display it.
1341 (thanks to Tobias C. Rittweiler)
1342 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1343 was able to infer the correct argument type for the object on which
1344 the CAS operation was being performed.
1345 * bug fix: copy propagation interfered with parallel assignment
1346 semantics in local calls. (reported by Paul Khuong)
1347 * bug fix: the signed modular fixnum shift compiled to wrong code on
1348 x86 and x86-64. (spotted by a slight modification to some of
1351 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1352 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1353 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1354 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1356 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1357 and generic function now print the object identity for anonymous
1358 instances. (thanks to Attila Lendvai)
1359 * enhancement: as an extension to MOP, SBCL now supports SETF of
1360 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1361 (thanks to Attila Lendvai)
1362 * enhancement: sb-cover has an alternate annotation mode which puts
1363 the form annotations at the car of the form, instead of annotating
1364 it completely. The benefit of this mode is that it shows better
1365 which forms were instrumented by the compiler.
1366 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1367 by printer control variables. (reported by Dan Corkill)
1368 * bug fix: system leaked memory when delayed output was performed by
1369 the OS in smaller chunks then expected. (thanks to David Smith)
1370 * bug fix: system leaked memory when file streams were not closed
1372 * bug fix: large objects written to slow streams that were modified
1373 after the write could end up with the modified state written to
1374 the underlying file descriptor.
1375 * bug fix: multiple threads operating in parallel on the same stream
1376 could cause buffer-overflows.
1377 * bug fix: source location information is stored correctly
1378 (broken since 1.0.6). This bug would generally show up as the
1379 Slime debugger highlighting the wrong form when the sldb-show-source
1381 * bug fix: Branch forms are again annotated as branches in the sb-cover
1383 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1384 on x86-64. (reported by Gregory Vanuxem)
1385 * bug fix: EQUALP could return wrong results for structures with raw
1386 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1387 word). (reported by Vjacheslav Fyodorov)
1388 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1391 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1392 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1393 atomic compare-and-swap operations on threaded platforms.
1394 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1395 allows assigning a global minimum value to optimization qualities
1396 (overriding proclamations and declarations).
1397 * enhancement: closed over variables can be stack-allocated on x86
1399 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1401 * optimization: slot definition lookup is now O(1). This speeds up
1402 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1403 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1404 * optimization: ASSOC and MEMBER can now be open-coded for all
1405 combinations of keyword arguments when second argument is constant
1406 and SPEED >= SPACE. In other cases a specialized version is
1408 * bug fix: using obsoleted structure instances with TYPEP and
1409 generic functions now signals a sensible error.
1410 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1411 (reported by Kristoffer Kvello)
1412 * bug fix: backtrace construction is now more careful when making
1413 lisp-objects from pointers on the stack, to avoid creating bogus
1414 objects that can be seen by the GC.
1415 * bug fix: defaulting of values in contexts expecting more than 7
1416 variables now works on x86-64. (reported by Christopher Laux)
1417 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1419 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1420 as the property-list of a symbol.
1421 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1422 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1423 situations requested, are once again file-compileable. (reported
1426 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1427 * MOP improvement: support for user-defined subclasses of
1428 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1429 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1430 the expansion of DEFMETHOD, is responsible for generating a form
1431 which creates a list of specializers when evaluated. Additional
1432 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1433 debugging and introspective support.
1434 * minor incompatible change: the (unsupported) spinlock interface
1435 has changed: free spinlock now has the value NIL, and a held spinlock
1436 has the owning thread as its value.
1437 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1438 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1440 * enhancement: name of a socket-stream is now "a socket" instead of
1441 "a constant string".
1442 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1443 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1444 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1445 for generic functions. (Thanks to Tobias C. Rittweiler)
1446 * optimization: bignum printing speed has been improved by 20-40%
1447 (depending on the bignum size.)
1448 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1450 * bug fix: the cache used by the CLOS to store precomputed effective
1451 methods, slot offsets, and constant return values is now thread and
1453 * bug fix: generic function dispatch function updating is now thread
1454 and interrupt safe (in the sense that the known issues have been
1456 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1457 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1458 * bug fix: asdf systems with dependencies to the SB-POSIX or
1459 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1460 * bug fix: interrupt safety of applicable method computation has been
1463 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1464 * new contrib: sb-cover, an experimental code coverage tool, is included
1465 as a contrib module.
1466 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1467 significantly faster.
1468 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1469 produce more efficient code on x86-64 if the compiler can determine
1470 that (+ INDEX <constant>) does not require a bounds check and FOO
1471 has an element type at least 8 bits wide.
1472 * enhancement: a new, experimental synchronous timeout facility is
1473 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1474 * enhancement: when a symbol name conflict error arises, the
1475 conflicting symbols are always printed with a package prefix.
1476 (thanks to Kevin Reid)
1477 * enhancement: stepping is now once again supported on the SPARC and
1478 MIPS platforms. (It is also now more likely to work on CheneyGC
1480 * enhancement: sb-sprof can now also track and report accurate call
1482 * bug fixes: the treatment of non-standard subclasses of
1483 SB-MOP:SPECIALIZER is more correct.
1484 * incompatible change: PURIFY no longer copies the data from the
1485 dynamic space into the static and read-only spaces on platforms
1486 that use the generational garbage collector
1487 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1489 * bug fix: GC race condition occasionally resulting in crashes with
1490 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1492 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1494 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1495 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1496 * bug fix: an asynchronous interrupt could previously leave the
1497 system running with GC inhibited.
1498 * bug fix: a DECLARE form evaluated at top-level now causes an error
1499 rather than silently (or verbosely) returning NIL.
1500 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1501 result in a error being signalled due to source form lookup errors.
1502 (reported by Peter Graves)
1504 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1505 * incompatible change: removed writer methods for host-ent-name,
1506 host-ent-addresses -- changing the values did not update the DNS
1508 * minor incompatible change: changed experimental JOIN-THREAD interface
1509 * documentation: the manual now lists reader and writer methods
1510 in class slot documentation sections. (thanks to Richard M Kreuter)
1511 * documentation: unwinding from asyncronous events has been
1512 documented as unsafe.
1513 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1514 in multithreaded application code.
1515 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1516 platforms. (thanks to James Anderson for the optimization hint)
1517 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1519 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1520 variants no longer cons.
1521 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1522 their NOT- variants no longer cons.
1523 * optimization: EQUAL hash tables no longer use SXHASH for objects
1524 of all data types, but instead use an EQL hash for types for which
1525 EQUAL is the same as EQL.
1526 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1527 are significantly faster.
1528 * optimization: new STRING-OUTPUT-STREAM implementation is much
1529 faster for large outputs and conses 30% less on average.
1530 * enhancement: XREF information is now collected for references made
1531 to global variables using SYMBOL-VALUE and a constant argument.
1532 * enhancement: SIGINT now causes a specific condition
1533 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1534 * bug fix: FILE-POSITION used to signal an error in some cases where
1535 ANSI requires it to return NIL.
1536 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1537 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1538 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1540 * bug fix: bad type declaration in the CLOS implementation has
1541 been fixed. (reported by James Anderson)
1542 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1543 (thanks to Lutz Euler)
1544 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1545 required by the CLHS. (reported by Samium Gromoff)
1546 * bug fix: dead unbound variable references now signal an error.
1547 (reported by Marco Monteiro)
1548 * bug fix: / with an unused value was being deleted in safe code.
1549 (thanks to Marco Monteiro and Kevin Reid)
1550 * bug fix: number of characters that can be written onto a single
1551 line in a file is unlimited.
1552 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1553 been fixed by inhibiting interrupts for when GC is disabled.
1554 * bug fix: some interrupt safety issues with GETHASH, (SETF
1555 GETHASH), CLRHASH and REMHASH have been fixed.
1556 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1557 specifier no longer causes infinite recursion.
1558 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1559 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1560 * bug fix: an &environment argument with the correct variable information
1561 is passed to macros that are expanded during byte compilation.
1562 (reported by Samium Gromoff)
1563 * bug fix: a logic error was causing the hash function for bignums to
1564 have excessive amounts of collisions. (reported by Faré Rideau)
1565 * bug fix: modifying the contents of an array could change the return
1566 value of SXHASH on that array, which is only allowed for strings
1567 and bit vectors. (bug introduced in 0.9.16)
1568 * improvement: the x86-64/darwin port now passes all tests (except
1569 for the debugger tests) but should still be considered
1570 experimental until this is fixed.
1571 * improvement: a style-warning is signaled for CASE (etc) clauses with
1572 duplicate keys. (thanks to Kevin Reid)
1573 * improvement: macos/x86 and macos/x86-64 now use mach exception
1574 handlers for dealing with illegal instructions (for trapping and
1575 error handling) and memory protection violations (for GC).
1577 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1578 * new platform: experimental support for x86-64/darwin (MacOS).
1579 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1580 and getnameinfo sockets functions are used instead of gethostbyaddr
1581 and gethostbyname, on platforms where the newer functions are available.
1582 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1584 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1585 don't print the startup banner, but behave as if --noinform was passed
1586 as a command line argument. (thanks to Kevin Reid)
1587 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1588 * optimization: code using alien values with undeclared types is much faster.
1589 * optimization: the compiler is now able to open code SEARCH in more cases.
1590 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1591 * bug fix: using standardized COMMON-LISP special variables as loop
1592 variables no longer signals bogus package lock violations. (reported
1594 * bug fix: declaring local loop variables to be of a range-limited type
1595 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1596 (reported by Andras Simon)
1597 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1598 bugs remain on x86-64.)
1599 * bug fix: the #= and ## reader macros now interact reasonably with
1600 funcallable instances.
1601 * bug fix: type-checks for function arguments were compiled using the
1602 compiler policy settings of the wrong lexical environment (in 1.0.2
1604 * bug fix: SHADOW accepts characters as string designators, as required
1605 by the spec (thanks to Eric Marsden)
1606 * bug fix: fixed GC safety issues when foreign functions are called with
1607 non-base strings as arguments
1608 * bug fix: more consistent error output for fatal-compiler-warnings, like
1610 * bug fix: functions with proper names were showing up as NIL in
1611 backtraces (reported by Edi Weitz, regression in 1.0.2)
1613 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1614 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1615 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1616 produce more efficient code on the x86 if the compiler can determine
1617 that (+ INDEX <constant>) does not require a bounds check and FOO
1618 has an element type at least 8 bits wide.
1619 * bug fix: references to undefined variables are handled the same way
1620 in toplevel forms as in the normal compiler
1621 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1623 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1624 (thanks to Jon Buller)
1625 * improvement: faster compilation times for complex functions
1626 * improvement: added readlink support to SB-POSIX (thanks to Richard
1629 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1630 * improvement: experimental support for mach exception handling on
1631 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1633 * improvement: support for GBK external format.
1634 (thanks to Chun Tian (binghe))
1635 * improvement: the debugger now displays variables that have been closed
1636 over, in code compiled with (DEBUG 2) or higher
1637 * improvement: support for executable cores on NetBSD (thanks to
1639 * new feature: added a RESTART-FRAME debugger command
1640 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1641 be used to provide an implementation for FILE-POSITION on Gray streams
1642 (thanks to Eric Marsden)
1643 * new feature: add syslog, openlog and closelog support to SB-POSIX
1644 (thanks to Richard Kreuter)
1645 * optimization: the function call overhead in code compiled with
1646 a high DEBUG optimization setting is significantly
1647 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1648 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1649 (thanks to Tony Martinez)
1650 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1651 evaluated once (reported by Kevin Reid)
1652 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1653 works again (thanks to Kilian Sprotte)
1654 * bug fix: an error is signaled for tagbodies with duplicate tags
1655 (thanks to Stephen Wilson)
1656 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1657 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1658 stack frames from alien callbacks.
1659 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1660 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1661 65k pages (thanks to David Woodhouse)
1662 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1664 changes in sbcl-1.0.1 relative to sbcl-1.0:
1665 * new platform: FreeBSD/x86-64, including support for threading.
1666 * new feature: the compiler stores cross-referencing information
1667 about function calls (who-calls), macroexpansion (who-macroexpands)
1668 and special variables (who-binds, who-sets, who-references) for code
1669 compiled with (< SPACE 3). This information is available through the
1670 sb-introspect contrib.
1671 * new feature: users may subclass SEQUENCE, and have instances of
1672 these classes interoperate with standard Common Lisp functions if
1673 a number of methods are defined. (This feature is experimental
1674 and the interface subject to change based on feedback from SBCL
1675 users and the general community)
1676 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1677 x86/x86-64, rather than the previous fixed depth of 8
1678 * improvement: another pthread back-end of mutex "pthread-futex".
1679 * bug fix: non-ascii command-line arguments are processed correctly
1680 (thanks to Yaroslav Kavenchuk)
1681 * bug fix: non-required arguments were not passed correctly when a method
1682 defined using DEFMETHOD was called from a mop-generated method using
1683 CALL-NEXT-METHOD (reported by Pascal Costanza)
1684 * bug fix: recursion is now permitted in accessors through
1685 SLOT-UNBOUND. (reported by Pascal Costanza)
1686 * bug fix: an error was signaled at startup if the HOME environment
1687 variable was defined, but had an empty value (reported by Peter Van Eynde)
1688 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1689 signal a PROGRAM-ERROR, not a BUG.
1690 * bug fix: SB-POSIX:READDIR works when built with large file support.
1691 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1693 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1694 for better automatic indentation support. (thanks to Matt Pillsbury)
1695 * optimization: loading generic functions no longer takes O(n^2) time,
1696 proportional to the amount of methods in the generic function
1697 (reported by Todd Sabin and Jeremy Brown)
1698 * optimization: the FIND and POSITION family of sequence functions
1699 are significantly faster on arrays whose element types have been
1701 * improvements to the Windows port:
1702 ** Intermittent heap corruption problems have been fixed. (thanks
1703 to Alastair Bridgewater)
1704 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1706 ** Lisp is able to unwind foreign exception frames from alien
1707 callbacks. (thanks to Alastair Bridgewater)
1709 changes in sbcl-1.0 relative to sbcl-0.9.18:
1710 * improvement: experimental support for threading on FreeBSD/x86.
1711 (thanks to NIIMI Satoshi)
1712 * improvement: runtime option --dynamic-space-size can be used
1713 to set the size of the dynamic space reserved on startup.
1714 * improvement: floating point modes in effect are now saved in
1715 core, and restored on startup.
1716 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1717 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1718 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1719 threads are not running after *SAVE-HOOKS* have run.
1720 * improvement: writes to CLOS instance slots are type-checked in code
1721 compiled with (SAFETY 3)
1722 * improvement: floating-point exception handling on FreeBSD (thanks to
1724 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1725 (thanks to Zach Beane)
1726 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1728 * improvement: added support for the Shift-JIS external format.
1729 (contributed by NIIMI Satoshi)
1730 * improvement: callbacks are supported on Linux/PPC. (thanks to
1732 * bug fix: compiler bug triggered by a (non-standard) VALUES
1733 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1734 * bug fix: file compiler no longer confuses validated and already
1735 dumped structurres. (reported by Kaersten Poeck)
1736 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1737 fill-pointer signals a type-error as required. (thanks to
1739 * bug fix: disassemly of funcallable instances works.
1740 * bug fix: single stepping on PPC.
1741 * bug fix: fix thread-safety problems in the type system (generally
1742 manifesting as nonsensical errors like "STRING is a bad type specifier
1743 for sequences" or "The value 1 is not of type FIXNUM")
1744 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1745 (reported by Josip Gracin)
1746 * bug fix: an error is signaled for attempts to displace arrays with
1747 incompatible element types (thanks to Mario Mommer)
1748 * bug fix: more correct handling of wide characters in debug info
1749 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1750 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1751 and don't cause extra consing
1752 * optimization: MAP and MAP-INTO are significantly faster on vectors
1753 whose elements types have been declared.
1754 * Improvements to SB-SPROF:
1755 ** Support for allocation profiling
1756 ** Reduced profiling overhead, especially for long profiling runs
1757 * Improvements to the Windows port:
1758 ** floating point exceptions are now reported correctly.
1759 ** stack exhaustion detection works partially.
1760 ** more accurate GET-INTERNAL-REAL-TIME.
1761 ** better RUN-PROGRAM behaviour in case of errors in the child
1763 ** PROBE-FILE now simplifies pathnames correctly.
1764 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1765 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1766 1st 1970 (although time timezone data will be unreliable).
1767 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1769 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1770 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1771 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1773 * bug fix: two potential GC deadlocks affecting threaded builds.
1774 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1775 returns T (reported by Anton Kazennikov)
1776 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1777 with non-variable places
1778 * bug fix: remove a race condition in the setting of
1779 funcallable-instance functions, this should make threaded CLOS
1780 code more stable against memory faults.
1781 * bug fix: corruption of specials when unbinding is interrupted by an
1782 asynchronous unwind (reported by Hannu Koivisto)
1783 * improvement: the debugger will now also display local variables that
1784 are only used once, for code compiled with a DEBUG optimization quality
1787 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1788 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1789 * incompatible change: External-format support for FFI calls. The
1790 SB-ALIEN:C-STRING no longer implies an ASCII
1791 external-format. Instead the string is subject to external-format
1792 conversion. Additionally return values of type SB-ALIEN:C-STRING
1793 are by default of type (SIMPLE-ARRAY CHARACTER), not
1794 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1795 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1796 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1797 * incompatible change: SB-EXT package no longer contains the
1798 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1799 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1801 * incompatible change: the single-stepper is no longer available
1802 on Alpha, Mips and Sparc platforms.
1803 * minor incompatible change: the direct superclasses of
1804 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1805 not (STANDARD-OBJECT FUNCTION). This makes the
1806 class-precedence-lists of GENERIC-FUNCTION and
1807 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1809 * minor incompatible change: the default stream external format on
1810 non-unicode SBCL is no longer detected from the locale, but is always
1812 * new feature: Add a version of evaluator that uses an interpreter instead
1813 of the compiler. EVAL still uses the compiler by default, to switch it
1814 to use the interpreter, set the value of the variable
1815 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1816 * minor incompatible change: the single-stepper REPL has been merged
1817 with the normal debugger (see the "Stepping" heading of the debugger help
1818 for more details). The debugger command STEP will no longer switch
1819 to the single-stepper REPL.
1820 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1821 for a type now works.
1822 * bug fix: loading of default sysinit file works. (thanks to Leonid
1824 * bug fix: better detection of circularities in the file-compiler.
1825 (reported by Marco Monteiro)
1826 * bug fix: the CL pathname functions now work with files that have
1827 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1828 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1829 systems with Unix98 pty semantics.
1830 * bug fix: ASDF-INSTALL will now work with bsd tar.
1831 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1833 * bug fix: timers expiring in dead threads no longer cause a
1834 type-error (reported by Paul "Nonny Mouse").
1835 * bug fix: thanks to more lightweight single-stepper instrumentation,
1836 code compiled with (DEBUG 3) will compile and execute significantly faster,
1837 and will have more accurate type-inferencing than before
1838 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1839 whose bindings are modified
1840 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1841 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1842 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1843 CL:SOFTWARE-VERSION with proper version
1845 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1846 * feature: implemented the READER-METHOD-CLASS and
1847 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1848 as specified by AMOP.
1849 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1851 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1852 (thanks to Lutz Euler)
1853 * optimization: hashing of general arrays and vectors has been
1854 improved. (reported by Any Fingerhut)
1855 * enhancement: SB-INTROSPECT is now able to find definitions of
1856 profiled functions. (thanks to Troels Henriksen)
1857 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1858 as well, allowing compiler-macros for SETF-functions to expand.
1859 * enhancement: step-instrumentation no longer wraps known
1860 single-value functions with multiple-value context, allowing
1861 better type inference.
1862 * fixed bug #337: use of MAKE-METHOD in method combination now works
1863 even in the presence of user-defined method classes. (reported by
1864 Bruno Haible and Pascal Costanza)
1865 * fixed bug #339(c): if there are applicable methods not part of any
1866 long-form method-combination group, call INVALID-METHOD-ERROR.
1867 (reported by Bruno Haible)
1868 * fixed bug #361: the :FUNCTION initarg in the protocol for
1869 initialization of methods can now be used to override
1870 internally-produced optimized functions. (reported by Bruno
1872 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1873 system-provided lambda expression no longer cause warnings about
1874 unbound #:|pv-table| symbols.
1875 * bug fix: improved the handling of type declarations and the
1876 detection of violations for keyword arguments with non-constant
1878 * bug fix: potentially erronous calls to PATHNAME and
1879 MERGE-PATHNAMES were being flushed in some cases.
1880 (reported by Richard Kreuter)
1881 * bug fix: compiled calls to TYPEP were mishandling obsolete
1882 instances. (reported by James Bielman and Attila Lendvai)
1883 * bug fix: format strings with ~> without matching ~< no longer
1884 trigger an AVER, but signal an understandable error instead.
1885 (reported by Antonio Martinez)
1886 * bug fix: specifying an output-file without a pathname-type for
1887 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1888 of input-file instead of "fasl". (reported by Robert Dodier)
1889 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1890 (reported by James Y Knight).
1891 * bug fix: compiler-macros-function did not consider the environment
1892 argument for shadowing by local functions.
1893 * bug fix: compiler-macros expansion was inhibited by local INLINE
1895 * bug fix: inline expansions of known functions were subject to
1896 step-instrumentation in high debug policies, leading to problems
1897 with type-inference.
1898 * bug fix: compiler failed to differentiate between different CONS
1899 types in some cases.
1900 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1901 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1902 * bug fix: #368: incorrect use of expressed vs. upgraded array
1904 * bug fix: #306a: more precise unions of array types.
1905 * thread-safety improvements:
1906 ** CONDITION-WAIT could return early on Linux, if the thread was
1907 interrupted and subsequently continued with SIGCONT.
1908 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1910 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1911 * added support for the ucs-2 external format. (contributed by Ivan
1913 * minor incompatible change: pretty printing of objects of type
1914 (cons symbol) is, in the default pprint-dispatch-table, now
1915 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1917 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1918 called later in a class's lifetime, possibly as late as when the
1919 first instance of the class is created. Previously,
1920 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1921 class became finalizeable.
1922 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1923 for multibyte external-format streams. (thanks to "vbzoli")
1924 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1925 with :ALLOCATION :CLASS inherited from superclasses of the
1927 * fixed bug: anonymous classes can now be created using the :NAME
1928 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1929 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1930 * fixed bug: core-files saved with :EXECUTABLE T can again be
1931 executed when SBCL_HOME isn't set. (reported by James Knight)
1932 * fixed bug: toplevel LOCALLY forms with declarations could
1933 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1934 * fixed bug: printing from several different threads using different
1935 values of *print-case* could cause invalid output, due to
1936 some internal special variables of the printer not being bound
1937 thread-locally (reported by Max Mikhanosha)
1938 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1939 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1940 (reported by Pascal Costanza)
1941 * fixed bug: FINALIZE-INHERITANCE is called from
1942 REINITIALIZE-INSTANCE on classes when the class has previously
1943 been finalized, as required by AMOP.
1944 * minor code generation optimizations:
1945 ** better register allocation in CLOS dispatching functions
1946 ** overflow detection when coercing signed bytes to fixnums on x86-64
1947 ** is now implemented with one IMUL instruction instead of three shifts
1948 ** more efficient bit-vector access on x86 and x86-64
1949 ** more efficient access to raw structure slots on x86 and x86-64
1950 * fixed some bugs revealed by Paul Dietz' test suite:
1951 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1952 return its argument.
1954 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1955 * feature: thread support on Solaris/x86, and experimental thread support
1957 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1959 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1960 default method for SLOT-UNBOUND.
1961 * minor incompatible change: prevent the user from specializing the
1962 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1963 somewhat counter to the intent of the protocol, I (CSR) think, and
1964 additionally it just doesn't work in SBCL as currently
1965 implemented, thanks to optimizations (that are always valid for
1966 the other three SLOT-VALUEish functions, but not for the setter).
1967 * minor incompatible change: the :SB-LDB feature is now enabled by
1968 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1969 the low-level debugger.
1970 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1971 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1972 down to an SB-SYS:FD-STREAM, allowing direct communication in
1973 more cases. Temporary files and pipes are still used as a fallback
1975 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1976 on the non-reentrant C functions localtime(3) and gmtime(3).
1977 * bug fix: global optimization policy was not visible in LOCALLY and
1979 * bug fix: class objects can be used as specializers in methods.
1980 (reported by Pascal Costanza)
1981 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1982 with a type error. (reported by blitz_ on #lisp)
1983 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1984 failed with a type error. (reported by Pascal Bourguignon)
1985 * bug fix: merging pathnames against defaults with :DIRECTORY
1986 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1987 (reported by James Y Knight)
1988 * bug fix: saving large (>2GB) cores on x86-64 now works
1989 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1990 constant index and a (simple-array (signed-byte 32)) array
1991 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1992 error (patch by Robert J. Macomber)
1993 * feature: TIME now displays the amount of run-time spent in GC
1994 * bug fix: The debugger now does a better job of respecting
1995 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1996 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1997 workaround for bug 403.)
1998 * bug fix: floating point exception handling now works on FreeBSD
1999 (thanks to NIIMI Satoshi)
2000 * fixed some bugs revealed by Paul Dietz' test suite:
2001 ** MISC.641: LET-conversion were not supposed to work in late
2003 * improvements to the Win32/x86 port:
2004 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
2005 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
2007 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
2008 * new feature: source path information is generated for macro-expansion
2009 errors for use in IDE's like Slime (thanks to Helmut Eller)
2010 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
2011 * bug fix: compiler does not loop forever on an invalid type in
2013 * improvement: compilation of most CLOS applications is significantly
2015 * optimization: added a limited bytecode compiler for simple toplevel
2016 forms, speeding up compilation and FASL loading
2017 * bug fix: the statistical profiler now properly distinguishes anonymous
2020 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
2021 * minor incompatible change: in sbcl-0.9.11 (but not earlier
2022 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
2023 into subclasses of the finalized class. Now user calls to
2024 FINALIZE-INHERITANCE finalize just the one class, and calls by the
2025 system return before any subclasses are finalized.
2026 * minor incompatible change: The reader no longer ignores errors
2027 regarding non-existent packages in #+ and #- feature tests.
2028 * new feature: command line options --no-sysinit, --no-userinit to
2029 inhibit loading the corresponding init files
2030 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
2031 for sending data through UDP sockets (thanks to François-René Rideau)
2032 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
2033 error is signalled instead (thanks to François-René Rideau)
2034 * minor incompatible change: Error signalling behaviour of lexical
2035 operations violating package locks has changed slightly. Refer to
2036 documentation on package locks for details.
2037 * enhancement: EVAL can process IF-expressions without resorting to the
2039 * bug fix: Some operations involving SIN, COS, or TAN weren't being
2040 constant-folded properly due to missing out-of-line functions.
2041 (reported by Mika Pihlajamäki)
2042 * bug fix: LISTEN sometimes returned T even in cases where no data was
2043 immediately available from the stream
2044 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
2045 were reversed. (reported by Levente Mészáros)
2046 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
2047 when given a vector displaced to an adjustable array. (reported
2049 * bug fix: derivation of float boundaries from numbers outside the
2050 appropriate float range (reported by John Wiseman)
2051 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
2053 * bug fix: precomputing cache entries for generic functions with
2054 some subclasses of specializers as yet invalid does not attempt to
2055 fill a cache line with a negative offset. (reported by Levente
2057 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
2058 allow condition class objects as arguments to DOCUMENTATION and
2059 (SETF DOCUMENTATION); only find and set documentation for
2060 structure names for the STRUCTURE doc-type. (suggested by Gary
2062 * improvements to the Win32/x86 port:
2063 ** (user-homedir-pathname) and default initialization file
2064 locations now know about the user's "Documents and Settings"
2065 directory (thanks to Yaroslav Kavenchuk)
2066 ** run-program is implemented (thanks to Mike Thomas)
2067 ** sockets support (thanks to Timothy Ritchey)
2068 ** better backtrace support (thanks to Alastair Bridgewater)
2069 ** sb-grovel supported
2070 ** asdf-install and sb-posix work somewhat
2071 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
2072 * improvements to the Solaris/x86 port:
2073 ** works on Solaris 11/Solaris Express
2074 ** floating-point exception handling support
2075 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
2076 * fixed some bugs revealed by Paul Dietz' test suite:
2077 ** REMOVE-METHOD returns its generic function argument even when
2078 the method is not one of the generic functions' methods.
2079 ** objects declared as MEMBER types can be admissible arguments to
2080 structure accessors.
2081 ** printing characters should simply be printed by the FORMAT ~:C
2083 ** compiler failure when compiling functions with hairy constant
2084 defaults for optional parameters.
2085 ** compiler produces wrong code when MAYBE-INLINE-expanding a
2086 function, which is already optimized.
2088 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
2089 * new platform: experimental support for SBCL x86/Darwin, including
2090 MacOS X 10.4.5 on Intel.
2091 * new feature: Unicode character names are now known to the system
2092 (through CHAR-NAME and NAME-CHAR).
2093 * new feature: the filesystem location of SBCL's core file is
2094 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
2095 * minor incompatible change: the contrib modules SB-POSIX and
2096 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
2097 this change is to make it easier to distribute
2098 location-independent binaries.
2099 * bug fix: as implied by AMOP, standardized classes no longer have
2100 slots named by external symbols of public packages. (reported by
2102 * optimization: calling functions via constant symbols -- as in
2103 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
2104 via the function object as in (FUNCALL #'FOO).
2105 * optimization: on x86 and x86-64, the calling convention for the
2106 case of dealing with an unknown number of values has been altered
2107 to be friendlier to the prediction heuristics implemented,
2108 particularly with reference to CALL and RET pairing. (thanks to
2109 Alastair Bridgewater)
2110 * optimization: on x86, the code for access to thread-local symbol
2111 values and binding/unbinding of thread-local symbols is smaller.
2112 (thanks to Alastair Bridgewater)
2113 * enhancement: CONSTANTP is now able to determine constantness of
2114 more complex forms, including calls to constant-foldable standardized
2115 functions and some special forms beyond QUOTE.
2116 * fixed bug: occasional GC crashes on Solaris/x86
2117 * optimization: x86-64 supports stack allocation of results of simple
2118 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
2119 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
2120 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
2121 (thanks to James Knight)
2122 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
2123 by the standard (thanks to Stephen Compall)
2125 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
2126 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
2127 be used for bundling the runtime and the core file into one
2128 executable binary. This feature is not currently supported on all SBCL
2129 platforms. (thanks to James Bielman and NIIMI Satoshi)
2130 * new feature: a generational or ephemeral garbage collector is now
2131 the default on the PowerPC platform (both Linux and Darwin). The
2132 old Cheney (stop and copy) collector is a build-time option.
2133 (thanks to Cyrus Harmon, after Raymond Toy)
2134 * minor incompatible change: the method by which SBCL finds its
2135 contributed modules has changed; it no longer relies on symbolic
2136 links from an $SBCL_HOME/systems directory, but searches directly
2137 in the subdirectories of $SBCL_HOME.
2138 * enhancement: the dynamic heap size on the Linux/PPC platform is
2139 markedly larger, even using the older Cheney garbage collector.
2140 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
2141 certain complicated string types. (reported by Gary King)
2142 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
2143 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
2144 * fixed bug: DOCUMENTATION on structures no longer signals an
2145 error if no structure type of the right name exists. (reported by
2147 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
2148 applicable non-standard primary or :AROUND method on
2149 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
2150 no longer cause unbound variable errors. (reported by Kevin Reid)
2151 * fixed bug: metacircle resolution in cases where methods have slots
2152 added before the slots from STANDARD-METHOD. (reported by Jean
2154 * fixed bug: the Power PC assembly code for calling into lisp
2155 sometimes computed its return address wrongly (depending on the
2156 15th bit of the address, fixed at link time). This used to
2157 manifest itself in a segmentation violation while building PCL.
2158 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
2159 many others over the years)
2160 * fixed bug: ROOM no longer reports silly numbers for stack usage if
2161 the stack pointer (treated as a signed integer) is negative.
2162 (thanks to Peter van Eynde)
2164 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
2165 * new platform: experimental support for the Windows operating
2166 system has been added. (thanks to Alastair Bridgewater)
2167 * fixed several bugs in and robustified the PPC FFI (including
2168 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
2169 * bug fix: allow non-simple string symbol names (reported by Paul
2171 * bug fix: interrupt handling on NetBSD (thanks to Richard M
2173 * bug fix: saving a core corrupted callbacks on x86/x86-64
2174 * bug fix: closed a loophole in metacircularity detection and
2175 grounding in the PCL implementation of CLOS.
2176 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2177 (reported by tomppa on #lisp)
2178 * bug fix: clear the direction flag on Lisp -> C transitions, as
2179 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2180 (reported by Andrej Grozin and Hendrik Maryns)
2181 * optimization: major improvements to GC efficiency on GENCGC platforms
2182 * optimization: faster implementation of EQUAL
2183 * optimization: emit more efficient opcodes for some common
2184 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2185 * optimization: several other minor code-generation improvements on x86-64
2186 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2188 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2189 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2190 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2191 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2192 by ANSI as a generic function, and for consistency (SETF
2193 GENERIC-FUNCTION-NAME) is treated likewise.
2194 * fixed bug #233.b: make constraint propagation notice when a variable
2195 value is changed after it is referenced but before it is used
2196 * fixed bug #296: no more arbitrary behaviour from filenames with
2197 odd characters as --load arguments. (reported by Adam Warner)
2198 * fixed bug #390: :CHARACTER-SET pathname components now work as
2199 expected. (reported by Tim Daly Jr)
2200 * fixed bug #391: complicated :TYPE intersections in slot
2201 definitions no longer cause an error in PCL internals.
2202 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2203 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2204 returning the number of octets which would be written to the
2205 file-stream. (thanks to Robert J. Macomber)
2206 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2207 arguments to the internal stream functions. (thanks to David
2209 * bug fix: stack exhaustion is now handled on the x86/SunOS
2210 platform. (thanks to Daisuke Homma)
2211 * bug fix: more accurate ROOM results on GENCGC platforms
2212 * optimization: improved type inference for arithmetic-for
2213 index variables in LOOP
2214 * optimization: faster floating-point SQRT on x86-64
2215 * fixed some bugs revealed by Paul Dietz' test suite:
2216 ** DOCUMENTATION returns NIL instead of "" for method combinations
2217 that don't have a docstring
2219 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2220 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2221 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2222 therefore are not customizeable by user code (as seems to be at
2223 least permitted and maybe required by AMOP). As a consolation,
2224 however, the SBCL implementation of these functions now calls
2225 REINITIALIZE-INSTANCE as specified by AMOP.
2226 * bug fix: slot-definition documentation is propagated as per ANSI
2227 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2228 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2229 argument, as do eslotd initialization methods. (from Pascal
2230 Costanza's "Closer" project)
2231 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2232 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2234 * bug fix: it is now possible to have more than one subclass of
2235 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2236 (reported by Bruno Haible, Pascal Costanza and others)
2237 * bug fix: the dependent update protocol now works for generic
2238 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2239 and Pascal Costanza)
2240 * bug fix: condition-class instances corresponding to
2241 DEFINE-CONDITION forms are now created eagerly. (reported by
2242 Kalle Olavi Niemitalo on comp.lang.lisp)
2243 * bug fix: floating point printing is more accurate in some
2244 circumstances. (thanks to Simon Alexander)
2245 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2246 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2247 * bug fix: callbacks on OS X now preserve stack-alignment.
2248 (thanks to Cyrus Harmon)
2249 * enhancement: the x86-64 disassembler is much better at
2250 disassembling SSE instructions. (thanks to Lutz Euler)
2251 * enhancement: CHECK-TYPE on a variable now allows type-inference
2252 (thanks to Vincent Arkesteijn)
2253 * optimization: improved performance of EUC-JP external format.
2254 (thanks to NIIMI Satoshi)
2255 * optimization: performance improvements to IO on file streams of
2256 :ELEMENT-TYPE CHARACTER
2257 * optimization: much faster memory allocation on x86-64
2258 * optimization: faster garbage collections (latency and throughput)
2260 * optimization: faster bignum subtraction and division on x86 and x86-64
2261 (thanks to Lutz Euler)
2262 * optimization: more accurate type-inference for loops that use a
2263 floating point index variable or a negative step.
2265 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2266 * bug fix: add a workaround to SBCL looping infinitely at startup on
2267 Linux kernels with apparently buggy implementations of personality().
2268 (thanks to Svein Ove Aas)
2269 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2270 (thanks to David Lichteblau)
2271 * bug fix: local bindings shadow global symbol macros for
2272 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2273 * optimization: non-open coded uses of numeric comparison operators
2274 (e.g. >) no longer cons when called with more than one parameter
2275 on platforms supporting dynamic-extent allocation.
2276 * enhancement: saving cores with foreign code loaded is now supported
2277 on MIPS/Linux in addition to the previously supported platforms.
2278 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2279 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2280 * bug fix: division by zero in sb-sprof when no samples were collected
2281 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2282 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2283 (thanks to Alexander Kjeldaas)
2284 * new feature: ignoring the return values of destructive functions
2285 like NREVERSE signals a compile-time style-warning.
2286 (thanks to Kevin Reid)
2287 * new feature: the alignment of alien structure fields can be
2288 explicitly specified. (thanks to Cyrus Harmon)
2289 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2290 and dump core on SIGQUIT
2292 ** incompatible change: threads do not inherit values of specials
2293 from their parents (see manual)
2294 ** bug fix: threads stacks belonging to dead threads are freed by the
2295 next exiting thread, no need to gc to collect thread stacks anymore
2296 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2297 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2298 ** bug fix: dynamic variable and thread start related gc lossage
2299 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2301 * fixed some bugs revealed by Paul Dietz' test suite:
2302 ** SUBTYPEP is slightly more accurate on heinously complicated
2303 CONS types where some of the members have uncertain (in the
2304 NIL, NIL sense) type relationships to each other.
2306 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2307 no longer purified unless :PURIFY T is explicitly specified.
2308 ** Non-purified cores are significantly smaller than before
2310 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2311 * new feature: timers based on Zach Beane's excellent timer package
2312 * added support for the following external formats: koi8-u,
2313 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2314 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2315 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2316 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2317 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2318 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2319 * incompatible change: a threaded SBCL will no longer revert to
2320 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2321 * bug fix: interrupts are disabled until startup is complete; no
2322 more sigsegvs when receiving a signal to soon
2323 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2325 * bug fix: add a workaround for the memory randomization features in
2326 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2327 workaround will only be in effect on systems with the proc filesystem
2329 * bug fix: printing objects of type HASH-TABLE signals a
2330 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2332 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2333 have been running for over 50 days. (reported by Gilbert Baumann)
2334 * bug fix: the logic for getting names of functions gets less
2335 confused when confronded with alternate-metaclass
2336 funcallable-instances. (reported by Cyrus Harmon)
2337 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2338 consistent, even on internal alternate-metaclass objects.
2339 * bug fix: flush closure information collected by physical
2340 environment analysis prepass before the main pass. (bug reported
2342 * bug fix: compiler pack phase does not modify a hashtable, which is
2343 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2344 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2345 may be expanded. (reported by Alexander <malishev> on c.l.l)
2346 * bug fix: no more style-warnings when compiling code using LAST
2347 under high-SPEED low-SPACE settings. (reported by David Wallin
2349 * fixed some incompatibilities between SBCL's MOP and the MOP
2351 ** the METAOBJECT class is now implemented;
2352 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2353 STANDARD-OBJECT, as required;
2354 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2355 now compatible (as required by VALIDATE-SUPERCLASS); there
2356 remains a consistency requirement over the presence or absence
2357 of the FUNCTION class in the superclass list of finalized
2358 classes; see the manual for more details;
2359 ** the :AROUND method for COMPUTE-SLOTS on
2360 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2361 requested slot ordering.
2363 ** bug fix: parent thread now can be gc'ed even with a live
2365 ** bug fix: binding a special with PROGV to no value is not
2366 the same as that symbol not having been bound (thanks to
2368 ** bug fix: binding specials is thread safe (thanks to
2370 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2371 and SB-SPROF do not die with 'no handler for signal XX in
2372 interrupt_handle_now(..)' anymore
2373 ** bug fix: WITH-TIMEOUT works with multiple running threads
2374 * fixed some bugs revealed by Paul Dietz' test suite:
2375 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2376 the :method-class keyword argument.
2378 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2379 * new port: the Solaris operating system on x86 processors is now
2380 mostly supported, though some rough edges in the environment
2381 remain. (thanks to Daisuke Homma)
2382 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2383 space for its heap. (thanks to Thiemo Seufer)
2384 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2385 closures. (thanks to Thiemo Seufer)
2386 * minor incompatible change: eof selects abort in the debugger.
2387 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2388 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2390 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2391 deprecated, and will go away in a future revision of SBCL.
2392 * minor incompatible change: GC-ON and GC-OFF are no longer
2393 implemented with a counter, it does not matter how many times gc
2394 is switched on or off
2395 * bug fix: discriminating functions for generic function classes
2396 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2397 make invalid assumptions about method precedence order. (reported
2399 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2400 platforms. (thanks to Thiemo Seufer)
2401 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2402 * bug fix: correct pathname printing with printer escaping is on.
2403 (thanks to Kevin Reid)
2404 * bug fix: complex VOP definitions in "user-space" no longer trigger
2405 package locks. (reported by Zach Beane)
2406 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2407 for funcallable-instance functions, and is no different from
2409 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2410 characters in the current readtable. (reported by Nicholas Neuss)
2411 * bug fix: SUBTYPEP on various CONS types returns more a more
2412 accurate acknowledgment of its certainty.
2413 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2414 lists in some cases. This partially fixes bug 384.
2415 * flush all standard streams before prompting in the REPL and the
2417 * bug fix: signal handling and triggering gc do not conflict
2418 directly anymore, in particular a high frequency sb-sprof does
2419 not prevent gc from running
2420 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2421 approximation for timezone and DST information between the
2422 universal time epoch and the smallest negative 32-bit time_t.
2423 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2424 year 1899 so long as the encoded time is non-negative.
2425 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2426 the correct answer for non-constant rotate amounts when performing
2427 an inline 32-bit rotation.
2429 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2430 there is only one thread in the session
2431 ** bug fix: memory leak for streams created in one thread and
2432 written to in another
2433 ** bug fix: lockup when compiled with gcc4
2434 ** bug fix: race that allows the gc to be triggered when gc is
2436 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2437 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2439 ** bug fix: finished threads are gc'ed properly
2440 * fixed some bugs revealed by Paul Dietz' test suite:
2441 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2442 has arguments with a different set of applicable methods from
2443 the orignal arguments.
2444 ** The type error thrown by MAP now has a correct expected-type
2446 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2447 name a compiled function.
2448 ** Three MISC tests where a large bignum was improperly coerced to
2449 a float that couldn't represent that bignum during type
2450 derivation were fixed.
2451 ** SUBTYPEP can now handle types involving the negation of a
2452 list-form FUNCTION type.
2453 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2454 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2455 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2457 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2458 * New feature: Experimental support for bivalent streams: streams
2459 opened with :element-type :default now allow character and binary
2460 (unsigned-byte 8) I/O
2461 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2462 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2464 * bug fix: no more highly sporadic "couldn't check whether ~S is
2465 readable" when reading a stream and an interrupt hits in the middle
2466 of a select system call
2467 * compiler better recognizes complex arrays (reported by Eduardo
2469 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2470 for variables on SBCL built with linkage-tables. (reported by Luis
2472 * various error reporting improvements.
2473 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2474 (Thanks to Thiemo Seufer)
2475 * bug fix: nested reader invokations work correctly
2476 * bug fix: it is possible to have simultaneous references to foreign
2477 code and foreign data with the same name.
2479 ** added x86-64 support
2480 ** incompatible change: the threading api now works with thread
2481 objects instead of thread ids
2482 ** bug fix: threads are protected from signals and interruption when
2483 starting up or going down
2484 ** bug fix: a race where an exiting thread could lose its stack to gc
2485 ** bug fix: don't halt on infinite error in threads if possible
2486 ** fixed numerous gc deadlocks introduced in the pthread merge
2487 ** bug fix: fixed thread safety issues in read and print
2488 ** bug fix: debugger doesn't hang on session lock if interrupted at
2489 an inappropriate moment
2490 ** bug fix: run-program is now thread safe(r)
2491 ** bug fix: inner with-recursive-lock no longer releases the mutex
2492 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2493 * fixed some bugs revealed by Paul Dietz' test suite:
2494 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2495 EXPECTED-TYPE slots filled.
2496 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2498 ** MAKE-CONDITION accepts classes as type-designators.
2499 ** COMPILE may never return NIL.
2500 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2501 range before calling Unix time functions
2503 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2504 * numerous signal handling fixes to increase stability
2505 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2506 * minor incompatible change: we now correctly canonize default
2507 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2508 as per the MOP, rather than the historical (INITARG INITFUNCTION
2509 INITFORM). (reported by Bruno Haible)
2510 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2511 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2512 for more information.
2513 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2514 pathname is a directory pathname.
2515 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2516 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2518 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2519 :METACLASS argument in addition to a class name. (reported by
2520 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2521 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2522 (reported by Bruno Haible; patch from Gerd Moellmann)
2523 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2525 * bug fix: more cleanups to the floating point exception handling on
2526 x86-64 (thanks to James Knight)
2527 * bug fix: the compiler does not try to constant fold calls of
2528 COERCE and BIT-* functions when they return freshly constructed
2529 objects. (reported by Nikodemus Siivola and Paul Dietz)
2530 * optimization: improved the MIPS versions of generic (in the
2531 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2532 * optimization: direct conversion of (unsigned-byte 32) to floats on
2533 the PowerPC platform.
2534 * optimization: structure instances with raw slots now use less
2535 memory, and probably show better memory locality. (thanks to
2537 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2538 are treated as requests for stack allocation on the x86-64,
2539 Alpha-32, PPC and SPARC platforms.
2540 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2541 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2542 (thanks to Tony Martinez)
2544 ** gcing a dead thread can no longer lead to lockups
2545 ** threads block signals until they are set up properly
2546 ** errno is no longer shared by threads
2547 ** interrupt-thread restores the eflags register on x86
2548 ** fixed some lockups due to gc/thread interaction
2549 * fixed some bugs revealed by Paul Dietz' test suite:
2550 ** invalid dotted lists no longer raise a read error when
2551 *READ-SUPPRESS* is T
2552 ** signal an error if a symbol that names a declaration is used
2553 as the name of a type, or vice versa
2554 ** allow using the (declare (typespec var*)) abbreviation for
2555 (declare (type typespec var*)) with all type specifiers
2556 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2557 ** always return NIL from PPRINT-POP when OBJECT is NIL
2558 ** don't signal errors when pretty-printing malformed LABELS,
2559 FLET or MACROLET forms
2560 ** declarations in a DOLIST body are also in scope for the
2562 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2564 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2567 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2568 * fixed cross-compiler leakages that prevented building a 32-bit
2569 target with a 64-bit host compiler.
2570 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2571 opened with :IF-EXISTS :OVERWRITE.
2572 * fixed bug 281: error for an invalid qualifier in a short-form method
2573 combination method is not signalled until the faulty method is called.
2574 * bug fix: iteration variable type inferrer failed to deal with open
2575 intervals. (reported by Alan Shields)
2576 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2577 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2578 or more runtime options were provided to the sbcl binary.
2579 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2580 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2581 * optimization: there's now a fast-path for fixnum arguments in the
2582 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2583 * optimization: the code generated on x86-64 is more compact thanks
2584 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2585 * bug fix: floating-point exception handling now partly works on
2586 x86-64. (Thanks to James Knight)
2587 * improvement to the MIPS backend from Thiemo Seufer: C-style
2588 64-bit long long arguments and return values to alien functions
2590 * fixed some bugs revealed by Paul Dietz' test suite:
2591 ** the type-error signalled from WARN has a filled-in DATUM slot.
2592 ** the type-error required when a stream is not associated with
2593 a file has the stream as its datum.
2594 ** type-errors on single-floats on x86-64 no longer have
2595 :INVALID-OBJECT as the datum
2596 ** the type-errors signalled for invalid function names now have
2597 a correct expected type
2598 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2599 for structures defined with a :TYPE. Documentation strings for
2600 typed structures are no longer immediately discarded
2601 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2602 broadcast streams and synonym streams. FILE-LENGTH now also works
2603 on broadcast streams.
2605 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2606 * incompatible change: the --noprogrammer option, deprecated since
2607 version 0.7.5, has been removed. Please use the equivalent
2608 --disable-debugger option instead.
2609 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2611 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2612 inoperational for a while now) has been completely removed.
2613 * null lexical environments are now printed as #<NULL-LEXENV>,
2614 significantly reducing the amount of clutter in typical backtraces.
2615 * documentation on weak pointers, finalization, and after GC hooks
2616 has been added to the manual.
2617 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2618 as other specialized array types, is much faster. SUBSEQ and
2619 COPY-SEQ on such arrays have also been sped up.
2620 * optimization: EQL is now more efficient when at least other argument
2621 is known to be of type (OR FIXNUM (NOT NUMBER)).
2622 * fixed bug: the runtime is now less vulnerable to changes in the
2623 size of the SBCL object on OS X, and virtual memory is reserved for
2624 all spaces that need to be at a fixed address.
2625 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2626 * fixed bug: finalizers and after GC hooks that cause consing are now
2628 * fixed bug: compiler error messages and summaries are now printed to
2629 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2630 * fixed inference of the upper bound of an iteration variable.
2631 (reported by Rajat Datta).
2632 * fixed bug 211e: calling local functions with duplicated constant
2633 keyword argument no longer causes a bogus style warning about an
2635 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2636 ftype declaration to be disregarded. (reported by Dave Roberts)
2637 * fixed bug 373: caused by erronous compilation of references to alien
2638 variables in the runtime on ppc/darwin.
2639 * fixed bug 376: CONJUGATE type deriver.
2640 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2641 (reported by Baughn on #lisp)
2642 * a cross-compiler bug on non-x86 platforms has been identified and
2643 fixed. (thanks to Bruno Haible)
2644 * improvements to the MIPS runtime code for increased stability.
2645 (thanks to Thiemo Seufer)
2646 * increased the maximimum compact environment size to allow
2647 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2648 * partial workaround for bug 135: don't name the function we're
2649 calling for hairy cases of EVAL, so as not to accumulate one environment
2650 entry per call to EVAL. (reported by Kevin Reid)
2651 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2652 * optimization: type testing for non-vector arrays should be faster.
2653 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2655 * bug fix: setting 31st element of a bit vector to zero did not work
2657 * bug fix: redefining a class definition which failed due to a
2658 previous accessor / function clash now works (but see BUGS entry
2659 #380 for more problems in this area). (thanks to Zach Beane)
2660 * the long form of DEFINE-METHOD-COMBINATION disables method group
2661 checking when given a single method group with pattern *. (thanks
2663 * on x86 compiler supports stack allocation of results of simple
2664 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2665 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2666 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2667 * fixed some bugs related to Unicode integration:
2668 ** the restarts for recovering from input and output encoding
2669 errors only appear when there is in fact such an error to
2671 * fixed some bugs revealed by Paul Dietz' test suite:
2672 ** MISC.549 and similar: late transformation of unsafe type
2673 assertions into derived types caused unexpected code
2675 ** SCALE-FLOAT type deriver is less wrong.
2676 ** type derivers for EXP, LOG and similar functions compute result
2677 types for complex arguments better.
2678 ** (MISC.563) CONJUGATE type deriver works for very restricted
2680 ** out-of-line type testers for character strings are available.
2681 ** EQUAL compiler transform understands specialness of objects
2683 ** accessing double-floats stored on the stack now works on x86-64.
2684 ** debugger internals could sometimes create invalid lispobjs,
2685 resulting in GC crashes.
2686 ** MISC.548: type check weakening can convert required type into
2688 ** initialization forms for bindings are not in scope of free special
2691 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2692 * incompatible change: thread support for non-NPTL systems has
2693 been removed - locking is buggy and unreliable. A threaded
2694 SBCL build will now warn at startup and refuse to create
2695 new threads, unless futex support is detected in the kernel
2696 * incompatible change: the top level REPL now has only an ABORT
2697 restart associated with it, not TOPLEVEL and ABORT as it used to.
2698 TOP and TOPLEVEL are now available as debugger commands for
2699 returning to the top level.
2700 * incompatible change: forms evaluated in the REPL now use the
2701 global optimization policy.
2702 * incompatible change: user- and system-initialization files are
2703 no longer processed with LOAD, but by READ and EVAL; hence the
2704 global optimization policy, startup package, readtable, etc,
2706 * The .fasl file format number has been incremented because of
2707 various incompatible changes.
2708 * internal entry point details and argument counts no longer appear
2709 in backtraces unless explicitly requested by setting
2710 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2711 * built-in and standard functions no longer have names like "top
2712 level local call to FOO".
2713 * fixed bug 32: functions defined in non-null lexical environments
2714 now have more legible printed representation
2715 * fixed bug 33: functions defined in non-null lexical environemnts
2716 are now more amenable to inspection by INSPECT.
2717 * workaround for bug 354: XEPs no longer appear in backtraces unless
2718 explicitly requested.
2719 * fixed bug: receiving the signal which results from attempting to
2720 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2721 SIGBUS on other *BSDs) on architectures where the C stack is also
2722 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2723 to debugger confusion or wild instability. (reported by Cheuksan
2725 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2726 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2727 (reported by Lutz Euler)
2728 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2729 compiled code. (reported by Paul Dietz)
2730 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2731 specialized parameter to a method to something that is not TYPEP
2732 the specializer is now possible.
2733 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2734 face of package deletion.
2735 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2736 pathname. (reported by David Tolpin on c.l.l)
2737 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2738 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2739 STRING-STREAM and FILE-STREAM, respectively.
2740 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2741 than previously; in addition, its test suite is now run on build.
2742 (thanks to Luke Gorrie)
2743 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2744 * optimization: added a immediate representation for single-floats
2746 * fixed some bugs revealed by Paul Dietz' test suite:
2747 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2748 ** attempting to create a package with a colliding nickname causes
2749 correctable errors to be signalled.
2750 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2751 ** several x86-64 backend bugs related to sign-extension of immediate
2754 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2755 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2756 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2758 * fixed loading of multiply forward-referenced layouts.
2759 (thanks to Cheuksan Wang)
2760 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2761 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2762 (see COMPILE-FILE documentation for details of :PRINT option)
2763 * fixed bugs 19 and 317: fixed-format floating point printing is
2764 more accurate. This also fixes a bug reported by Adam Warner
2765 related to the ~@F format directive.
2766 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2768 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2769 dispatch table if the from-char is a dispatch macro character.
2770 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2771 by the wordsize no longer ignore the last word. (reported by Lutz
2773 * fixed bug in type checking of dynamic-extent variables. (reported
2775 * optimization: sequence traversal functions use their freedom to
2776 coerce function designators to functions.
2777 * optimization: code with many calls to CLOS methods specialized on
2778 CLOS classes has had redundant type checks removed.
2779 * optimization: added declarations to speed up operations that access
2780 the internal character database (for example STRING-UPCASE)
2781 * optimization: comparison operations between floats and sufficiently small
2782 fixnums no longer create extra rationals
2783 * fixed some bugs related to Unicode integration:
2784 ** portions of multibyte characters at the end of buffers for
2785 character-based file input are correctly transferred to the
2786 start of the buffer at the next read.
2787 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2788 passing it through to OPEN.
2789 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2790 argument given to internal calls to OPEN.
2791 ** the built-in comment readers (introduced by character sequences
2792 ";" and "#|") are more forgiving to encoding errors; they will
2793 STYLE-WARN and then attempt to resync the stream at a character
2794 boundary. (thanks to Teemu Kalvas)
2795 * fixed some bugs in the x86-64 port:
2796 ** Negative short int return values from c-calls are sign-extended
2798 ** The stack is aligned to 16-bytes for c-calls, as required by
2799 the ABI. (thanks to Cheuksan Wang)
2800 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2801 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2802 for enums. (thanks to Vincent Arkesteijn)
2803 ** Multiple small optimizations and bugfixes for floating point
2805 * fixed some bugs revealed by Paul Dietz' test suite:
2806 ** Space, Tab, Linefeed, Return and Page have the invalid
2807 secondary constituent character trait.
2808 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2810 ** WITH-INPUT-FROM-STRING should only update the index place on
2812 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2813 works more reliably.
2814 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2815 with one-dimensional arrays with fill pointers.
2816 ** TYPE-OF failed on a complex with an integer realpart and a
2818 ** compiler failure during type inference for the code of form
2819 (IF (EQL X (THE ...)) ...) (MISC.535).
2821 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2822 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2823 platform. The port supports 61-bit fixnums, large memory spaces
2824 and reloading shared object files.
2825 * enhancement: saving cores with foreign code loaded is now
2826 supported on ppc/Darwin in addition to the previously supported
2828 * enhancement: the statistical profiler now walks deeper into the
2829 call stack for more meaningful call-graphs and accrued time
2830 reports (x86/x86-64 only). It also now reports time spent in
2832 * enhancement: it is now possible to trace most individual methods
2833 of a generic function in addition to tracing the generic function
2835 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2837 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2838 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2839 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2840 produces an error. (thanks to Vincent Arkesteijn)
2841 * bug fix: NAMESTRING on pathnames with :WILD components in their
2842 directories works correctly. (thanks to Artem V. Andreev)
2843 * fixed bug 125: compiler preserves identity of closures. (reported
2845 * bug fixed: functions with &REST arguments sometimes failed with
2846 "Undefined function" when compiled with (DEBUG 3). (reported by
2848 * bug fix: overflow during compiling of setting element of a bit
2849 vector with constant index and value. (reported by Timmy Douglas)
2850 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2851 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2852 * on x86 compiler partially supports stack allocation of dynamic-extent
2854 * GO and RETURN-FROM do not check the extent of their exit points
2855 when compiled with SAFETY 0.
2856 * fixed some bugs related to Unicode integration:
2857 ** encoding and decoding errors are now much more robustly
2858 handled; it should now be possible to recover even from invalid
2859 input or output to the terminal. (thanks to Teemu Kalvas)
2860 ** provided a first cut at implementing STRING-TO-OCTETS and
2861 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2862 ** altered the SB-MD5 contributed module slightly, changing the
2863 interface just enough for it to be supportable for builds where
2864 lisp characters are not eight bits.
2865 * fixed some bugs revealed by Paul Dietz' test suite:
2866 ** the FORMATTER-generated functions for ~V[ conditionals require
2867 the correct number of arguments.
2868 ** READ-FROM-STRING returns the mandated second value when applied
2869 to displaced strings.
2870 ** the #\Rubout and #\Backspace characters are treated as invalid
2871 constituent characters by the tokenizer.
2873 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2874 * new feature: reloading changed shared object files with
2875 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2876 * new feature: references to foreign variables and functions
2877 can now be compiled and loaded before the corresponding shared
2878 object file is loaded, as long as the foreign definitions are
2879 available at runtime.
2880 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2881 supported. (thanks to Dan Debertin)
2882 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2883 just keywords. (thanks to Vincent Arkesteijn)
2884 * fixed bug #331: structure-class instances corresponding to
2885 DEFSTRUCT forms are now created eagerly.
2886 * fixed bug #345: backtraces from calls to undefined functions work
2887 on x86 as well. Related bug #61 is now also partially fixed on x86
2888 (backtraces from throws to unknown catch tags.)
2889 * bug fix: lambda-list parsing is now stricter vrt. order and number
2890 of lambda-list keywords.
2891 * bug fix: as specified by AMOP, an error is signalled if a
2892 class-option appears multiple times in a DEFCLASS form. (reported
2894 * bug fix: RANDOM can be compiled when the compiler derives the type
2895 of its numeric argument as a disjoint set of small integers.
2896 (reported by Paul Dietz)
2897 * bug fix: starting a core saved with shared objects loaded when
2898 those objects are not available no longer causes threaded SBCL to
2899 hang. (reported by Sean Ross)
2900 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2902 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2903 argument, which must always be NIL. (reported by Kalle Niemitalo)
2904 * bug fix: printing 1.0d+23 no longer results in an error.
2905 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2906 * bug fix: structure slot setters preserve evaluation order. (thanks
2908 * fixed some bugs related to Unicode integration:
2909 ** RUN-PROGRAM can allow its child to take input from a Lisp
2910 stream. (reported by Stefan Scholl)
2911 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2912 locales. (reported by Ken Causey)
2913 * fixed some bugs revealed by Paul Dietz' test suite:
2914 ** INCF, DECF and REMF evaluate their place form as specified in
2916 ** FORMATTER expands ~{ ~} iteration directives with V or #
2917 parameters correctly.
2918 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2919 consequent uses no arguments correctly.
2920 ** the system has a partial understanding of the (COMPLEX RATIO)
2922 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2925 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2926 * new feature: a build-time option (controlled by the :SB-UNICODE
2927 keyword feature, enabled by default) for building the system with
2928 support for the entire 21-bit character space defined by the
2930 * new feature: the system now has rudimentary external-format
2931 support; the primary user-visible change at this time is that
2932 characters with the high bit set (such as lower-case-e-acute) will
2933 print correctly to a terminal in a UTF-8 environment.
2934 * minor incompatible change: BASE-CHAR no longer names a class;
2935 however, CHARACTER continues to do so, as required by ANSI.
2936 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2937 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2938 has been moved to the SB-EXT package (temporarily re-exported from
2940 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2941 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2943 * fixed bug #351: better error-handlind and reporting for malformed
2945 * fixed bug #350: bignum-printing is now more memory-efficient,
2946 allowing printing of very large bignums, eg. (expt 2 10000000).
2947 (reported by Bruno Haible)
2948 * fixed bug #302: better primitive-type selection for intersection
2950 * fixed bug #308: non-graphic characters now all have names, as
2951 required. (reported by Bruno Haible)
2952 * bug fix: redefining a class with different superclasses now correctly
2953 removes it from the direct-subclasses of its previous superclasses.
2954 (reported by David Morse)
2955 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2956 new value now works. (reported by Bruno Haible)
2957 * bug fix: correct canonicalization of multiple non-standard slot
2958 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2959 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2960 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2961 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2962 now exists, an signals an error.
2963 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2964 during macroexpansion for non-iteration variables as well. (reported
2965 by Bruno Haible for CMUCL)
2966 * bug fix: Cyclic structures and unprintable objects in compiler
2967 messages no longer cause errors. (reported by Bruno Haible)
2968 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2969 and READ-FROM-STRING all now return a primary value of NIL if
2970 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2971 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2972 (reported by Bruno Haible for CMUCL)
2973 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2974 do not designate a condition. (reported by Bruno Haible for
2976 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2977 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2978 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2979 permitting resolution in favour of any of the conflicting symbols.
2980 (reported by Bruno Haible for CMUCL)
2981 * bug fix: EQUAL compiler optimizations is less aggressive on
2982 strings which can potentially compare true despite having distinct
2983 specialized array element types.
2984 * bug fix: unit enumerations can be defined without dividing by
2985 zero. (thanks to Vincent Arkesteijn)
2986 * FORMAT compile-time argument count checking has been enhanced.
2987 (report from Bruno Haible for CMUCL)
2988 * a partial workaround for the bug 262: the compiler does not try to
2989 inline-expand a local function doing RETURN-FROM from a deleted
2990 BLOCK. (thanks to Peter Denno for the bug report and to David
2991 Wragg for the simple test case)
2992 * fixed some bugs revealed by Paul Dietz' test suite:
2993 ** RENAME-PACKAGE allows all package designators as new package
2995 ** constraint propagation and lambda variable substitution are
2996 more cautious in dealing with partially deleted code.
2997 ** compiler sometimes lost reoptimization passes.
2998 ** CERROR, given a condition as condition designator, passes its
2999 remaining arguments to the continue format control without
3001 ** Case-altering FORMAT directives work correctly on non-ASCII
3003 ** The REAL type specifier handles bounds outside the floating
3004 point ranges without signalling FLOATING-POINT-OVERFLOW.
3005 ** Functions with IR1-transformations can create intercomponent
3006 references to global functions.
3007 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
3009 ** FORMAT ~R treats a nil value for its first parameter correctly.
3011 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
3012 * enhancement: saving cores with foreign code loaded is now
3013 supported on x86/NetBSD and sparc/Linux in addition to the previously
3014 supported platforms.
3015 * bug fix: on some platforms repeated installations caused multiple
3016 copies of HTML documentation to be installed -- should not happen
3017 any more. (reported by Stefan Scholl)
3018 * bug fix: parsing self-recursive alien record types multiple times
3019 no longer causes infinite recursion. (reported by Thomas F. Burdick,
3020 original patch by Helmut Eller for CMUCL)
3021 * bug fix: stack-exhaustion detection works now on NetBSD as well.
3022 (thanks to Richard Kreuter)
3023 * bug fix: defining classes whose accessors are methods on existing
3024 generic functions in other (locked) packages no longer signals
3025 bogus package lock violations. (reported by François-René Rideau)
3026 * bug fix: special variables as DEFMETHOD parameters no longer have
3027 associated bogus type declarations. (reported by David Wragg and
3029 * bug fix: read-write consistency on streams of element-type
3030 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
3031 * bug fix: redefiniton of the only method of a generic function with
3032 no DEFGENERIC no longer emits a full WARNING. In addition,
3033 redefinition of generic functions with no DEFGENERIC to an
3034 incompatible lambda list now signals an error. (thanks to Zach
3036 * bug fix: DEFGENERIC now works even when there's a function of the
3037 same name in an enclosing lexical environment. (thanks to Zach
3039 * fixed compiler failure, caused by instrumenting code during
3040 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
3041 * optimization: added loop analysis and improved register allocation
3042 * fixed some bugs revealed by Paul Dietz' test suite:
3043 ** POSITION on displaced vectors with non-zero displacement
3044 returns the right answer.
3045 ** (SIMPLE-STRING) is a valid type specifier for sequence
3047 ** *PRINT-LEVEL* handling for slotless structures is pedantically
3049 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
3050 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
3052 ** The justification version of the FORMAT ~< directive treats
3053 non-zero minpad parameter correctly.
3055 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
3056 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
3057 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
3058 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
3059 the supported interface.
3060 * new feature: Single-stepping of code compiled with DEBUG 2 or
3061 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
3062 * new feature: saving cores with foreign code loaded is now
3063 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
3064 Timothy Moore's work for CMUCL)
3065 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
3066 parameters to * instead of NIL if no initform is supplied.
3067 (reported by Johan Bockgaard)
3068 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
3069 correctly when FUNCALL appears as the car of the form. Note:
3070 despite this FUNCALL forms are not currently subject to
3071 compiler-macro expansion. (port of Raymond Toy's fix for the
3072 same from CMUCL, reported by Johan Bockgaard)
3073 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
3074 (thanks for Teemu Kalvas)
3075 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
3076 case instead of terminating the iteration (thanks for Julian
3077 Squires, Sean Champ and Raymond Toy)
3078 * bug fix: incorrect expansion of defgeneric that caused a style
3079 warning. (thanks for Zach Beane)
3080 * bug fix: slot accessor effective method computation works properly
3081 for classes with multiple non-standard applicable methods on
3082 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
3083 * on x86 compiler supports stack allocation of results of LIST and
3084 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
3085 CMUCL implementation by Gerd Moellmann)
3086 * fixed some bugs revealed by Paul Dietz' test suite:
3087 ** FORMAT strings with both the ~<~:;~> form of the justification
3088 directive and pretty-printing directives cause an error.
3090 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
3091 * incompatible change: the internal functions
3092 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
3093 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
3094 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
3095 instead of the old functions.
3096 * new feature: on platforms where "dladdr" is available foreign
3097 function names now appear in backtraces. (based on Helmut Eller's
3099 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
3100 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
3102 * documentation: ASDF manual is now distributed with SBCL.
3103 * bug fix: SBCL can now load its contributed modules with REQUIRE
3104 even if the system-provided entries have been removed from
3105 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
3107 * bug fix: backtraces involving undefined functions or assembly
3108 routines are more informative. (thanks to Brian Downing)
3109 * bug fix: mutually referent alien structures now work correctly.
3110 (reported by Rick Taube)
3111 * bug fix: structures defined by WITH-ALIEN can be referred to
3112 within other definitions in the same WITH-ALIEN.
3113 * bug fix: division operators (MOD, TRUNCATE and the like) with
3114 constant zero divisors and integer dividends no longer generate
3116 * bug fix: provide default methods for INPUT-STREAM-P and
3117 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
3118 * bug fix: improve the detection and resolution of MOP metacycles
3119 (where effective method computation depends on the generic function
3120 having its effective method computed). (reported by Bruno Haible)
3121 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
3122 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
3123 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
3124 * bug fix: pathnames with relative directory components can be
3125 represented relative to default pathnames.
3126 * optimization: in taking the GCD of bignums, reduce the two bignums
3127 to approximately the same size (using Euclid's algorithm) before
3128 applying the more sophisticated binary GCD. (thanks to Juho
3130 * optimization: COUNT on bitvectors now operates word-at-a-time.
3131 * optimization: ASH with a positive, but not necessarily constant,
3132 (leftwards) shift, when in a modular context, is compiled to a
3134 * fixed some bugs revealed by Paul Dietz' test suite:
3135 ** FORMAT variable parameters ("~V<char>") are defaulted properly
3136 if the corresponding argument is NIL.
3137 ** FORMAT directives accepting a minpad parameter treat negative
3139 ** literal commas as character parameters to FORMAT directives are
3141 ** literal spaces directly after ~<Newline> directives within a
3142 format-logical-block (~:< ~@:>) do not induce :FILL-style
3143 conditional newlines.
3144 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
3145 as specified: it no longer includes conditional newlines).
3146 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
3148 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
3149 :PER-LINE-PREFIX argument does not evaluate to a string.
3151 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
3152 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3153 the manual for details; for now, package locks can be disabled by
3154 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
3155 if no major problems are found then it is likely that they will be
3156 compiled in unconditionally.
3157 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
3158 unsupported operators on all platforms. To load a shared library
3159 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
3160 object file, link it into a shared library outside of SBCL and
3161 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
3163 * bug workaround: ROOM T and the GC were not working together
3164 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
3165 reporting facility aren't preserved. That reporting has been
3166 disabled, so now until and unless someone figures out how to make it
3167 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
3168 * minor incompatible change: as threatened around sbcl-0.8.0, the
3169 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
3170 The symbols are also exported from SB-PCL for backwards
3171 compatibility, but more so than before SB-PCL should be treated as
3172 an implementation-internal package.
3173 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3175 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3176 It is known as of this release not to work on the Alpha, however.
3177 * fixed bug #167: errors signalled due to illegal syntax in method
3178 bodies are now more legible.
3179 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3180 designators and can hence be used with TYPEP.
3181 * fixed bug #333: CHECK-TYPE now ensures that the type error
3182 signalled, if any, has the right object to be accessed by
3183 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3184 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3185 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3187 * fixed bug #334: programmatic addition of slots using specialized
3188 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3189 and :ALLOCATION :CLASS slots.
3190 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3191 float exponents. (rereported by Peter Seibel)
3192 * fixed a bug: #\Space (and other whitespace characters) are no
3193 longer considered to be macro characters in standard syntax by
3194 GET-MACRO-CHARACTER.
3195 * fixed bug: initialization of condition class metaobjects no longer
3196 causes an instance of the condition to be created. (reported by
3198 * fixed bug: it is now possible to have slots such that
3199 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3200 description is neither :INSTANCE nor :CLASS.
3201 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3202 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3203 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3204 non-standard SLOT-DEFINITION-ALLOCATION values.
3205 * fixed bugs: various race conditions handling exiting threads.
3206 CL-PPCRE's thread torture test now passes.
3207 * fixed arguably-a-bug: GC time proportional to number of threads in
3208 system even when most of them are idle
3209 * optimization: improved performance of BIT and SBIT on bit-vectors.
3210 * .fasl file incompatibility: The fasl file version number has
3211 been incremented because of changes associated with package locks.
3213 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3214 * minor incompatible change: the system no longer provides
3215 optimization hints (conditions which are TYPEP
3216 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3217 mode; these hints are still emitted when higher SPEED optimization
3219 * new contrib module: a sampling profiler (profiling by statistical
3220 sampling, rather than by instrumenting functions) is available as
3221 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3222 * the behaviour of the standard function ED is now customizeable by
3223 third parties through a hook variable: see ED's documentation
3224 string for information on the protocol.
3225 * the compiler no longer emits efficiency notes for (FUNCALL X)
3226 when the type of X is uncertain under default optimization
3228 * fixed bug 276: mutating a binding of a specialized parameter to a
3229 method to something that is not TYPEP the specializer is now
3231 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3232 exhibit the required behaviour on the x86 platform. (thanks to
3233 Peter van Eynde, Eric Marsden and Bruno Haible)
3234 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3235 even for difficult arguments. (reported by Peter Graves)
3236 * fixed bug 141a: the backquote printer now descends quoted
3238 * fixed another bug in backquote printing: no more destructive
3239 modification of the form's list structure. (reported by Brian
3241 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3242 move between its address being taken and the call to
3243 interrupt_thread, fixing a crashing race condition.
3244 * the SB-POSIX contrib implementation has been adjusted so that it
3245 no longer exhibits ridiculously poor performance when constructing
3246 instances corresponding to C structs.
3248 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3249 * minor incompatible change: the sb-grovel contrib now treats C
3250 structures as alien (in the sense of SB-ALIEN) objects rather than
3251 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3252 has implications for memory management of client code
3253 (sb-grovel-returned objects must now be manually managed) and for
3254 type safety (alien objects now have full types).
3255 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3256 used to control emission of compiler diagnostics, rather than the
3257 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3258 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3259 quality should be considered deprecated.
3260 * install.sh now installs the user manual as well
3261 * (not quite a new documentable feature, but worth considering in
3262 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3263 beginnings of a semantically meaningful condition hierarchy is
3264 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3266 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3267 designator as the defaults argument.
3268 * fixed bug: Displaced arrays whose displaced-to array has become
3269 too small now cause ARRAY-DIMENSION to signal an error, providing
3270 for safer bounds-checking. (reported by Bruno Haible)
3271 * fixed bug: DEFCLASS slot definitions with identical :READER and
3272 :WRITER names now signal a reasonable error. (reported by Thomas
3274 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3276 * fixed bug: Invalid slot specification errors now print correctly.
3277 (thanks to Zach Beane)
3278 * fixed bug 320: Shared to local slot value transfers in class
3279 redefinitions now happen corrently from superclasses as
3280 well. (reported by Bruno Haible)
3281 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3283 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3284 improper lists correctly. (reported by Bruno Haible)
3285 * fixed bug 313: source-transform for <fun-name> was erroneously
3286 applied to a call of a value of a variable with name <fun-name>.
3287 (reported by Antonio Menezes Leitao)
3288 * fixed bug 307: The obsolete instance protocol ensures that
3289 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3290 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3291 remove dead unknown-values globs from the middle of the stack.
3292 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3293 easier to resume long computations after using *BREAK-ON-SIGNALS*
3294 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3295 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3296 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3297 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3298 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3299 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3301 * fixed a (fairly theoretical) bug in string printing: if
3302 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3303 does not have array-element-type equal to the most general string
3305 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3306 function name. (thanks to Nikodemus Siivola)
3307 * fixed bug: compile-time format string checker failed on
3308 non-closed ~{. (reported by Thomas F Burdick)
3309 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3310 don't warn on legal loop code involving more than one aggregate
3311 boolean. (thanks to Nikodemus Siivola)
3312 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3313 in keyword position to MAKE-PACKAGE was not being reported as
3315 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3316 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3318 * fixed bugs 280 and 312: the checking for multiple definitions in a
3319 file is less likely to become confused by uses of inline
3321 * fixed bug: the #S reader macro performs the keyword coercion
3322 specified for slot names. (reported by Kalle Niemitalo)
3323 * fixed bug: lambda lists may contain symbols whose names start with
3324 & but are not lambda-list-keywords; their occurrence triggers a
3326 * fixed bug 321: define-method-combination argument lambda lists do
3327 not cause invalid code to be generated when &optional or &aux
3328 variables have default values. (reported by Bruno Haible)
3329 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3330 have CLOS classes; this appears also to have fixed reported
3331 instability in stack exhaustion detection.
3332 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3333 MUFFLE-WARNING when no associated restart is present is now
3335 * optimization: rearranged the expansion of various defining macros
3336 so that each expands into only one top-level form in a
3337 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3339 * optimization: used a previously little-used slot in symbols to
3340 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3342 * fixed some bugs revealed by Paul Dietz' test suite:
3343 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3345 ** Fixed an optimization bug related to inheritance of initforms
3346 from local to shared slots.
3347 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3348 ** CAST optimizer forgot to flush argument derived type.
3349 ** print/read consistency on floats is now orders of magnitude
3350 more likely. (thanks also to Bruno Haible for a similar report
3352 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3353 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3354 for objects of type REAL. Make it so.
3355 ** SXHASH is specified (infelicitously) to respect similarity,
3356 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3357 it so. (thanks to Markus Ziegler)
3358 ** on the Alpha, the compiler succeeds in compiling functions
3359 returning a known number of arguments greater than 63.
3360 ** fixed handling of invalid NIL arguments in keyword position
3362 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3364 ** provided a readably-printable representation for RANDOM-STATE
3366 ** ensured that pathnames, where they have a namestring, always
3367 print using #P"..." syntax.
3369 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3370 * Support for the forthcoming 2.0 version of the NetBSD kernel
3371 running on x86 hardware has been added. (thanks to Perry
3372 E. Metzger most immediately, and others for their past work)
3373 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3374 current release version; SBCL's OpenBSD support had been broken
3375 since about the time of OpenBSD's switch to ELF binary format.)
3376 * [placeholder for DX summary]
3377 ** user code with &REST lists declared dynamic-extent, under high
3378 speed or space and low safety and debug optimization policy.
3379 * The manual has been converted to Texinfo format and the debugger
3380 chapter from the cmucl manual has been added.
3381 * A facility has been added to extract documentation strings from
3382 sbcl and store them as Texinfo-formatted snippets for inclusion in
3383 the manual (via Texinfo's @include directive)
3384 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3385 the test case to Dave Roberts)
3386 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3387 pointers. (reported by Sean Ross)
3388 * bug fix: PROFILE output is printed nicely even for large numerical
3389 values. (thanks to Zach Beane)
3390 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3391 greater than 32 handle EOF correctly.
3392 * bug fix: on X86 an immediate argument of the IMUL instruction is
3393 correctly printed in disassembly. (thanks to Lutz Euler)
3394 * bug fix: class slots in redefined classes preserve their old
3395 values. (thanks to Bruno Haible and Nikodemus Siivola)
3396 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3397 succeeds. (reported by Marco Baringer)
3398 * fixed some bugs revealed by Paul Dietz' test suite:
3399 ** READ-SEQUENCE now works on ECHO-STREAMs.
3400 ** RATIONALIZE works more according to its specification. (thanks
3403 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3404 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3405 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3406 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3407 should matter to you only if you rebind the printer control
3408 variables and then find you want different bindings in the
3409 debugger than in the ordinary execution of your program.)
3410 * The runtime build system has been tweaked to support building
3411 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3412 assembler and linker. (thanks to Nikodemus Siivola)
3413 * Unbound, undefined, undeclared variables now trigger full
3414 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3415 more often programmer error than deliberate exploitation of undefined
3417 * optimization: the hash algorithm for strings has changed to one
3418 that is less vulnerable to spurious collisions. (thanks to Juho
3420 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3421 needless bounds checking. (thanks to Juho Snellman)
3422 * optimization: implemented multiplication as a modular
3423 (UNSIGNED-BYTE 32) operation on the PPC backend.
3424 * fixed some bugs revealed by Paul Dietz' test suite:
3425 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3427 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3428 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3429 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3431 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3433 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3435 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3436 their output stream on EOF from read.
3437 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3438 have been read to end-of-file.
3439 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3441 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3442 description of determination of which consecutive characters
3444 ** Printing the "Space" character with escaping on now yields "#\\ ",
3445 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3446 ** Reading floating-point numbers with *READ-BASE* set to a number
3447 less than 10 works correctly.
3448 ** Reading floating-point numbers with *READ-BASE* set to a number
3449 more than 10 works correctly.
3450 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3451 the readtable currently in effect.
3453 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3454 * minor incompatible change: parsing of namestrings on a physical
3455 (Unix) host has changed; numbers after the final #\. in a
3456 namestring are no longer interpreted as a version field. This is
3457 intented to be largely invisible to the user, except that the
3458 meaning of the namestring "*.*.*" has changed: it now refers to a
3459 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3460 should usually be replaced by
3461 (make-pathname :name :wild :type :wild :version :wild)
3462 with the added benefit that this is more likely to be portable.
3463 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3464 option to OPEN now signals an error if the file being opened
3465 exists; this may have an impact on existing code.
3466 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3467 signals received, on the PowerPC platforms (both Linux and
3468 Darwin). (thanks to Pierre Mai for pointing out the location of
3470 * several fixes on OS X: The system now builds and runs cleanly on
3471 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3472 (thanks to Brian Mastenbrook)
3473 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3474 second-resolution: integer multiples of 1/3600 between -24 and 24.
3475 (thanks to Vincent Arkesteijn)
3476 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3477 type when called with 1 argument; PEEK-CHAR checked type of
3478 PEEK-TYPE only after having read first character from a
3479 stream. (reported by Peter Graves)
3480 * bug fix: the garbage collector now has much better locality
3481 behaviour, and in particular no longer treats all memory as being
3482 exhausted when many small objects point to each other in a deeply
3484 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3485 recognized as being TYPEP their class.
3486 * bug fix: the PUSHNEW documentation string has been corrected.
3487 (thanks to Vincent Arkesteijn)
3488 * bug fix: defaulting of the value for the last of an atypically
3489 large number of multiple values being bound was not being
3490 performed correctly on the Alpha or PPC platforms
3491 * optimization: implemented multiplication as a modular
3492 (UNSIGNED-BYTE 32) operation on the x86 backend.
3493 * optimization: SEARCH on simple-base-strings can now be open-coded.
3494 (see also contrib/compiler-extras.lisp for inspiration for
3495 teaching the compiler about the Boyer-Moore algorithm).
3496 * value, returned by MAX (and MIN) called with several EQUALP, but
3497 not EQL, arguments now does not depend on compiler settings.
3498 * fixed some bugs revealed by Paul Dietz' test suite:
3499 ** in stack analysis liveness information is propagated from
3500 non-local entry points.
3501 ** pathwise CAST removing failed when the CAST node did not start
3503 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3504 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3506 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3507 host is already defined.
3508 ** RENAME-FILE works on streams instead of signalling an internal
3510 ** PEEK-CHAR uses the current readtable when determining whether
3511 or not a character is whitespace.
3512 ** MERGE-PATHNAMES handles the case when the pathname does not
3513 specify a name while the default-pathname specifies a version.
3514 ** Pathnames now stand a better chance of respecting print/read
3516 ** Attempting to use standardized file system operators with a
3517 pathname with invalid :DIRECTORY components signals a
3519 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3521 ** DIRECTORY on logical pathnames is more correct.
3522 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3523 signal a TYPE-ERROR if their argument is not a stream.
3524 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3525 designator argument does not designate a stream.
3526 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3527 examining the synonym.
3528 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3530 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3531 element-type larger than ([UN]SIGNED-BYTE 32).
3533 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3534 * When built with the :SB-FUTEX feature, threaded builds now take
3535 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3536 for faster/more reliable mutex and condition variable support.
3537 * Incompatible change (but one you probably shouldn't have been using
3538 anyway): the interface and code for arbitrating between multiple
3539 threads in the same user session has been redesigned.
3540 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3541 error when called without an explicit environment argument.
3542 (thanks to Nikodemus Siivola)
3543 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3544 confused about their position. (thanks to Adam Warner and Gerd
3546 * bug fix: Pretty printing backquoted forms with unquotations in the
3547 argument list position of various code constructs such as LAMBDA
3548 now works correctly. (reported by Paul Dietz)
3549 * bug fix: Pretty printing unquotations no longer loses all
3550 stream position information.
3551 * optimization: performance of string output streams is now less
3552 poor for multiple small sequence writes.
3553 * optimization: performance of CSUBTYPEP in the presence of complex
3554 expressions involving CONS and NOT many times has been improved.
3555 (reported by Paul Dietz)
3556 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3558 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3559 streams for the REPL, for the debugger, and so on. (thanks to
3561 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3562 * fixed some bugs revealed by Paul Dietz' test suite:
3563 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3564 is ignored if it is too small, rather than propagating through
3565 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3566 ** extremely complex negations of CONS types were not being
3567 sufficiently canonicalized, leading to inconsistencies in
3569 ** VALUES tranformer lost derived type.
3571 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3572 * fixed a bootstrapping bug: the build process no longer assumes
3573 that the various BOOLE-related constants have the same value in
3574 host and target lisps. (noted by Paul Dietz' test suite on an
3575 SBCL binary built from CLISP)
3576 * The system can now be dynamically linked on the MIPS platform,
3577 which enables dynamic loading of foreign code from Lisp. (thanks
3578 to Ralf Baechle for discussions on the MIPS ABI)
3579 * The system now records debugging information for its own source
3580 files in a filesystem-position-independent manner, relative to
3581 the "SYS" logical host.
3582 * fixed a compiler bug: MV-LET convertion did not check references
3583 to the "max args" entry point. (reported by Brian Downing)
3584 * tweaked disassembly notes to be less confident about proclaiming
3585 some instruction as an LRA. (thanks to Brian Downing)
3586 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3587 now each have their own history, command character, and other
3588 characteristics. (thanks to David Lichteblau)
3589 * fixed some bugs revealed by Paul Dietz' test suite:
3590 ** compiler failure in compiling LOGAND expressions including a
3592 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3593 shift greater than 32.
3594 ** FLUSH-DEST did not mark blocks for type check regeneration.
3595 ** HANDLER-CASE failed to accept declarations in handler clauses
3596 in some circumstances.
3598 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3599 * New code in contrib/sb-introspect (still probably not entirely
3600 stable yet) provides some support for smart Lisp development
3601 environments like SLIME.
3602 * The conditions signalled for errors occurring when loading .fasl
3603 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3604 in a way which should help ASDF recover gracefully.
3605 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3606 stuff has been cleaned up. If you code contrib/ stuff, this might
3607 affect you, and you can look at contrib/README, contrib/STANDARDS,
3608 and/or the 0.8.4.27 diff to check.
3609 * In full calls the compiler now does not generate checks for declared
3610 argument types for all arguments.
3611 * various threading fixes
3612 ** and some experimental patches which didn't make it into
3613 the main tree for this release, but which are shipped in
3614 contrib/experimental-thread.patch as a possible fix for some
3615 failures (deadlock, spinning...) in GC-intensive multithreaded
3617 * fixed PPC build problem (source code incompatibility of different
3618 library versions): added offsetof() hackery which attempts to divine
3619 where glibc maintainers put uc_mcontext today
3620 * fixed bug 282: compiler does not trust type assertions while passing
3621 arguments to a full call.
3622 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3623 &OPTIONAL VALUES type parameter.
3624 * fix bug 214: algorithm for noting rejected templates is now more
3625 similar to that of template seletion. (also reported by rydis on
3627 * fixed bug 141b: printing backquoted information readably and prettily
3628 inserts a space where necessary.
3629 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3630 (CHARACTER 10) are now reported as errors, rather than propagated
3631 as unknown types. (reported by piso on #lisp)
3632 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3633 with values NIL and :ERROR. (thanks to Milan Zamazal)
3634 * fixed bug 191c: CLOS now does proper keyword argument checking as
3635 described in CLHS 7.6.5 and 7.6.5.1.
3636 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3637 longer raise an error; further, using a list as a for-as-arithmetic
3638 counter now raises a meaningful error.
3639 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3640 now understood by sequence creation functions such as MAKE-SEQUENCE
3642 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3643 a STRING-INPUT-STREAM.
3644 * compiler enhancement: SIGNUM is now better able to derive the type
3646 * type declarations inside WITH-SLOTS are checked. (reported by
3648 * fixed some bugs revealed by Paul Dietz' test suite:
3649 ** incorrect optimization of TRUNCATE for a positive first
3650 argument and negative second.
3651 ** compiler failure in let-convertion during flushing dead code.
3652 ** compiler failure while deriving type of TRUNCATE on an
3653 interval, containing 0.
3654 ** ASH of a negative bignum by a negative bignum count now returns
3656 ** intersection of CONS types now canonicalizes properly, fixing
3657 inconsistencies in SUBTYPEP.
3659 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3660 * incompatible change: The --disable-debugger command line
3661 option now clobbers the debugger at a more fundamental
3662 level, by redefining #'INVOKE-DEBUGGER instead of by
3663 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3664 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3665 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3666 enter the debugger REPL and then suffer recursive errors
3667 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3668 while under the new system, BREAK in --disable-debugger mode
3669 terminates the system just as an unhandled error would.
3670 * fixed compiler performance when processing loops with a step >1;
3671 * bug fix: DOCUMENTATION now retrieves generic function
3672 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3673 support has been systematized, and now supports the methods
3674 specified by ANSI, along with a default method and a method for
3675 slot documentation. (reported by Nathan Froyd)
3676 * bug fix: effective methods associated with a generic function are
3677 no longer cached over a change of that generic function's method
3678 combination. (reported by Andreas Fuchs)
3679 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3680 names a non-existent directory. (thanks to Andreas Fuchs)
3681 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3682 the PPC platform, lead to stack corruption; this has been fixed.
3683 (reported by Rainer Joswig)
3684 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3685 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3686 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3687 arithmetic in general, is now much more reliable on the PPC
3689 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3690 platform now returns the right answer.
3691 * optimization: restored some effective method precomputation in
3692 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3693 precomputation is now tunable.
3694 * optimization: compiler-internal data structure use has been
3695 reviewed, and changes have been made that should improve the
3696 performance of the compiler by about 20%.
3697 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3698 simple-base-strings and simple-bit-vectors is improved.
3699 * optimization: the optimization of 32-bit logical and arithmetic
3700 functions introduced in version 0.8.3 on the x86 has been
3701 implemented on the mips, ppc and sparc platforms; an
3702 implementation of the same facility, but for 64-bit arithmetic,
3703 has been added for the alpha.
3704 * microoptimization: the compiler is better able to make use of the
3705 x86 LEA instruction for multiplication by constants.
3706 * bug fix: in some situations compiler did not report usage of
3707 generic arithmetic in (SPEED 3) policy.
3708 * bug 145b fix: compiler used wrong type specifier while converting
3709 MEMBER-types to numeric.
3710 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3712 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3713 index argument now works properly on the Alpha platform.
3714 * bug fix: floating point exception treatment on the Alpha platform
3716 * bug fix: FILE-POSITION works much better on string input and
3717 output streams. (thanks to Nikodemus Siivola)
3718 * bug fix: many threading/garbage collection symptoms sorted.
3719 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3720 might be pseudo-atomic.
3721 * internal change: Stopping for GC is now done with signals not
3722 ptrace. GC is now done in whichever thread wanted it, instead of
3724 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3726 * bug fix: result form in DO is not contained in the implicit
3728 * incompatible change: ICR structure is changed; the value part of
3729 CONTINUATION is now called LVAR; corresponding functions are
3730 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3731 * added type deriver for ISQRT (thanks to Robert E. Brown).
3732 * bug fix: better support for loading from the command line when an
3733 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3735 * fixed some bugs revealed by Paul Dietz' test suite:
3736 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3737 ** ROUND and FROUND now give the right answer when given very
3738 small float arguments.
3739 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3741 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3742 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3743 ** LCM should return a non-negative integer.
3744 ** PARSE-INTEGER returned the index of a terminator instead of the
3745 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3746 ** PARSE-INTEGER returned an incorrect index being applied to a
3748 ** LCM with two arguments of 0 returns 0 rather than signalling
3750 ** unsigned addition of a 32-bit constant with the high bit set no
3751 longer causes an internal compiler error on the x86.
3752 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3753 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3754 with negative last argument.
3755 ** byte specifiers with zero size and position no longer cause
3756 an error during type derivation.
3757 ** bignum multiplication on the Alpha platform now returns the
3759 * porting: The system now builds on SuSE AMD64, although it still
3760 generates a 32-bit binary.
3761 * .fasl file incompatibility: The fasl file version number has
3762 been incremented (because of the changes to internal compiler
3763 data structures referred to above).
3765 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3766 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3767 more accurately, on the Darwin kernel running on PowerPC hardware.
3768 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3769 * Compiler code deletion notes now signal a condition of type
3770 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3771 an associated MUFFLE-WARNING restart.
3772 * The compiler now performs limited argument count validation of
3773 constant format strings in FORMAT, and where appropriate in ERROR,
3774 CERROR and WARN. (thanks to Gerd Moellmann)
3775 * New ASDF-INSTALL contrib can be used for automatic download and
3776 installation of third-party Lisp code from CCLAN or other sites
3778 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3779 which forces another thread to execute a function supplied by the
3781 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3782 now accept and act upon their :ELEMENT-TYPE keyword argument.
3783 (reported by Martin Atzmueller, Edi Weitz)
3784 * bug fix: FILE-POSITION now accepts position designators up to
3785 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3786 is the greater. (thanks to Patrik Nordebo)
3787 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3788 Kalle Olavi Niemitalo)
3789 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3790 printable. (reported by Eric Marsden)
3791 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3792 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3793 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3794 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3795 (thanks to Stig Sandoe)
3796 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3797 UNREAD-CHAR now correctly includes the unread character in the
3798 target sequence. (thanks to Gerd Moellmann)
3799 * bug fix in threaded builds: the system can now be suspended and
3800 resumed by shell job control with minimal disruption.
3801 * bug fixes in times and timezones >2038AD
3802 * better handling of "where is GNU make?" problem in build scripts
3803 (thanks to Nikodemus Siivola)
3804 * new optimization: inside a named function any reference to a
3805 function with the same name is considered to be a self-reference;
3806 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3807 optimization quality.
3808 * new optimization on x86: logical functions and + now have
3809 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3810 used when the result is truncated to 32 bits.
3811 * VALUES declaration is partially enabled.
3812 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3813 * fixed some bugs revealed by Paul Dietz' test suite:
3814 ** The system now obeys the constraint imposed by
3815 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3816 types form a lattice under type intersection.
3817 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3818 ** ASSOC now ignores NIL elements in an alist.
3819 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3820 and (1+ MOST-POSITIVE-FIXNUM) answers.
3821 ** The addition of a method with invalid qualifiers to a generic
3822 function does not cause an error to be signalled immediately;
3823 a warning is signalled, and the error is generated only on
3824 calling the generic function.
3825 * changed .fasl file version number, in part to add type codes for
3826 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3827 obscure ANSI requirements
3829 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3830 * fixed bug 148: failure to inline-expand a local function left
3831 garbage, confusing the compiler.
3832 * fixed bugs 3cd: structure slot readers perform type check if the
3833 slot can have an invalid value (i.e. it is either not initialized
3834 or can be written with a less specific slot writer).
3835 * bug fix: the compiler now traps array references to elements off
3836 the end of an array; previously, the bounds checking in some
3837 circumstances could go off-by-one.
3838 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3840 * type declarations for array element types now obey the description
3841 on the CLHS page "Declaration TYPE", as per discussions on
3842 sbcl-help around 2003-05-08. This means that a declaration
3843 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3844 declaration, all references to BAR will be asserted or assumed
3845 (with THE, so dependent on compiler policy) to involve objects of
3846 type FOO. Note that no such declaration is implied in
3847 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3848 * declared types of functions from the "Conditions"
3849 chapter. (reported by Paul Dietz)
3850 * bug fix: CERROR accepts a function as its first argument.
3851 * bug fix: NTH an NTHCDR accept a bignum as index
3852 arguments. (reported by Adam Warner)
3853 * optimization: character compare routines now optimize comparing
3854 against a constant character. (reported by Gilbert Baumann)
3855 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3856 argument now works properly on the MIPS platform.
3857 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3858 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3859 declared to have a functional type, does not cause an error
3861 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3862 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3863 * fixed bug: if last continuation of a deleted block has a
3864 destination, this destination should be deleted too. (reported by
3866 * fixed a bug in the bootstrap process: the host compiler's values
3867 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3868 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3869 test case from Patrik Nordebo)
3870 * improved the ability of the disassembler on the PPC platform to
3871 provide helpful disassembly notes.
3872 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3873 the class in more cases than previously.
3874 * bug fix: FILE-POSITION now understands :START and :END for
3875 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3876 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3877 8) data. (thanks to David Lichteblau)
3878 * bug fix: it is possible to add a method to a generic function
3879 without lambda list.
3880 * bug fix: reader failed to signal END-OF-FILE inside an
3881 object representation. (reported by Nikodemus Siivola)
3882 * fixed some bugs revealed by Paul Dietz' test suite:
3883 ** LAST and [N]BUTLAST should accept a bignum.
3884 ** condition slot accessors are methods.
3885 ** (VECTOR NIL) is a subtype of STRING.
3887 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3888 * minor incompatible change: some nonsensical specialized lambda
3889 lists (used in DEFMETHOD) which were previously ignored now signal
3891 * minor incompatible change: the system is now aware of the types of
3892 variables in the COMMON-LISP package, and will signal errors for
3893 most violations of these type constraints (where previously they
3894 were silently accepted).
3895 * minor incompatible change: COMPILE-FILE now uses the freedom
3896 afforded (ANSI 3.2.2.3) to use derived function types for
3897 functions defined in the same file. This also permits the system
3898 to warn on static type mismatches and function
3899 redefinition. (Currently it does not work with high DEBUG level.)
3900 * minor incompatible change: VALUES declaration is disabled.
3901 * When issuing notes, the compiler now signals a condition of type
3902 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3903 restart for use in user handlers. It is expected that the
3904 COMPILER-NOTE condition will eventually become a condition
3905 supertype to a hierarchy of note types, which will then be
3906 handleable in a similar fashion. However, at the moment, no such
3907 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3908 but it's an implementation detail, not a classification for the
3910 * Changes in type checking closed the following bugs:
3911 ** type checking of unused values (192b, 194d, 203);
3912 ** template selection based on unsafe type assertions (192c, 236);
3913 ** type checking in branches (194bc).
3914 * A short form of VALUES type specifier has ANSI meaning (it has
3915 increased the number of situations when SBCL cannot perform type
3917 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3918 keywords or constants is permissible.
3919 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3920 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3921 classes. (thanks to Antonio Martinez)
3922 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3923 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3924 argument to a pretty-print stream. Instead, it leaves any such
3925 operations to DESCRIBE-OBJECT methods.
3926 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3928 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3929 lambda lists. (thanks to David Lichteblau)
3930 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3931 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3932 (thanks to Gerd Moellmann)
3933 * fixed bug 246: increased compilation speed of long
3934 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3936 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3938 * DEFCONSTANT now throws a condition of type
3939 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3940 constant to a non-EQL value; CONTINUE and ABORT restarts
3941 respectively change and preserve the value.
3942 * fixed bug 63: The code walker, part of the implementation of CLOS,
3943 is now better at handling symbol macros.
3944 * bug fix: There is no longer an internal implementation type named
3945 CL:LENGTH. (reported by Raymond Toy)
3946 * bug fix: In macro-like defining macros/special operators the
3947 implicit block does not enclose the lambda list.
3948 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3949 enclosed in parentheses) are not suitable as type specifiers, and
3950 their use properly signals an error now.
3951 * bug fix: An argument count mismatch for a type specifier in code
3952 being compiled no longer causes an unhandled error at compile
3953 time, but signals a compile-time warning.
3954 * fixed simple vector readable printing
3955 * bug fix: DESCRIBE takes more care over whether the class
3956 precedence list slot of a class is bound before accessing it.
3957 (reported by Markus Krummenacker)
3958 * bug fix: FORMATTER can successfully compile pretty-printer format
3959 strings which use variants of the ~* directive inside.
3960 * bug fix: SEARCH now applies its TEST predicate to the elements of
3961 the arguments in the correct order. (thanks to Wolfhard Buss)
3962 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3963 in type inference. (thanks to Robert E. Brown)
3964 * bug fix: Reading in symbols with an explicit package name of ""
3965 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3966 * fixed some bugs revealed by Paul Dietz' test suite:
3967 ** NIL is now allowed as a structure slot name.
3968 ** Arbitrary numbers, not just REALs, are allowed in certain
3969 circumstances in LOOP for-as-arithmetic clauses.
3970 ** Multiple class redefinitions before slot access no longer
3971 causes a type error.
3972 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3973 association between the name and a class.
3974 ** Generic functions with non-standard method-combination and over
3975 five methods all of which return constants no longer return NIL
3976 after the first few invocations. (thanks to Gerd Moellmann)
3977 ** CALL-NEXT-METHOD with no arguments now passes the original
3978 values of the arguments, even in the presence of assignment.
3979 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3981 ** DISASSEMBLE works with closures and funcallable instances.
3982 ** ADD-METHOD now returns the generic function, not the new method.
3983 ** FIND-METHOD signals an error if the lengths of the specializers
3984 is incompatible with the generic function, even if the ERRORP
3986 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3987 which its argument is a member.
3988 ** DEFCLASS only redefines the class named by its class-name
3989 argument if that name is the proper name of the class;
3990 otherwise, it creates a new class.
3991 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3992 of the UNBOUND-SLOT condition to the name of the slot.
3993 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3994 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3995 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3996 SLOT-MAKUNBOUND in the specified fashion.
3998 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3999 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
4000 cross-compilation host. As a consequence, we can now bootstrap our
4001 way up to SBCL starting with a bare gcc toolchain and human-readable
4002 source code (first the source to CLISP, then the source to SBCL).
4003 * A contributed module containing a partial implementation of the
4004 simple-streams interface has been included. (thanks to Rudi
4006 * A contributed module implementing the RFC1321 Message Digest
4007 Algorithm, known as MD5, has been included.
4008 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
4009 no longer has any effect, as the code controlled by this feature
4010 has been deleted. (As far as we know, no-one has ever built using
4011 this feature, and its semantics were confused in any case).
4012 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
4013 work on conditions (as required by the ANSI specification),
4014 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
4015 expected behaviour on conditions. Users should note, however,
4016 that such behaviour is not required by the ANSI specification,
4017 and so use of this behaviour may render their code unportable.
4018 * fixed some bugs revealed by Paul Dietz' test suite:
4019 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
4021 ** &ENVIRONMENT parameter in macro lambda list is bound first.
4022 ** SXHASH on condition objects no longer returns NIL.
4023 ** :ALLOCATION :CLASS slots are better treated; their values are
4024 updated on class redefinition, and initforms inherited from
4025 superclasses are applied.
4026 ** REMOVE-METHOD returns its generic function argument even when
4027 no method was removed.
4028 ** SHARED-INITIALIZE now initializes the values of the requested
4029 slots, including those with :ALLOCATION :CLASS.
4030 ** ALLOCATE-INSTANCE now works on structure classes defined via
4031 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
4033 ** SLOT-EXISTS-P now works on conditions, as well as structures
4035 ** MAKE-LOAD-FORM now has the required methods on
4036 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
4037 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
4038 keyword, and now implements the SLOT-NAMES argument.
4039 ** methods with &OPTIONAL arguments no longer allow too many
4040 arguments to be passed in the call without error.
4041 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
4042 option is consistent with the required arguments of the generic
4043 function lambda list.
4044 * bug fix: REQUIRE accepts a string designator. (Thanks to
4046 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
4047 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
4048 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
4049 * bug fix: The compiler now checks for duplicated variables in macro
4051 * bug fix: SETQ on globals returns the correct value.
4052 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
4053 not a COMPILER-ERROR (followed by some other strange error on
4054 choosing the CONTINUE restart).
4055 * bug fix: make.sh and friends are now more consistent in the way that
4056 they look for GNU "make".
4058 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
4059 * experimental native threads support (on x86 Linux >=2.4 only).
4060 This is not compiled in by default: you need to add :SB-THREAD to
4061 the target features. See the "Beyond ANSI" chapter of the manual
4063 * fix for longstanding nonANSIism: The old distinction between
4064 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
4065 The return value from CL:FIND-CLASS is now a CLOS class, and
4066 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
4067 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
4068 * An interface to the MetaObject Protocol, as described in Kiczales,
4069 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
4070 MIT Press, 1991, is available from the SB-MOP package.
4071 * incompatible change: the SB-PCL package should now be considered
4072 a private implementation detail, and no longer a semi-private MOP
4074 * minor incompatible change: due to rearrangement for threads, the
4075 control stack and binding stack are now allocated at arbitrary
4076 addresses instead of being hardcoded per-port. Users affected by
4077 this probably have to be doing advanced things with shared
4078 libraries, and will know who they are.
4079 * minor incompatible change: Previously, all --eval forms used were
4080 processed with READ before any of them were processed with EVAL.
4081 Now each --eval form is processed with both READ and EVAL before
4082 the next --eval form is processed. (Thus package operations like
4083 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
4084 work as the user might reasonably expect.)
4085 * minor incompatible change: *STANDARD-INPUT* is now only an
4086 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
4088 * minor incompatible change: Y-OR-N-P is now character-oriented,
4089 not line oriented. Also, YES-OR-NO-P now works without errors.
4090 (thanks to Antonio Martinez)
4091 * sb-aclrepl module improvements: an integrated inspector, added
4092 repl features, and a bug fix to :trace command.
4093 * Known functions, which cannot be open coded by the backend, are
4094 considered to be able to check types of their arguments. (fixing
4095 a bug report by Nathan J. Froyd)
4096 * fixed a bug in computing method discriminating functions: It is
4097 now possible to define methods specialized on classes which have
4098 forward-referenced superclasses. (thanks to Gerd Moellmann)
4099 * fixed evaluation order in optional entries (reported by Gilbert
4101 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
4102 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
4103 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
4104 required-by-AMOP NAME argument, as well as CLASS and
4105 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
4106 * fixed bug 20: DEFMETHOD can define methods using names that are
4107 not the proper names of classes to designate class specializers.
4108 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
4109 Unix file descriptors, instead of blowing up. (thanks to Antonio
4111 * Garbage collection refactoring: user-visible change is that a
4112 call to the GC function during WITHOUT-GCING will not do garbage
4113 collection until the end of the WITHOUT-GCING. If you were doing
4114 this you were probably losing anyway.
4115 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
4116 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
4117 TYPEP the latter but not the former.
4118 * The compiler issues a full WARNING for calls to undefined functions
4119 with names from the CL package.
4120 * MAP-INTO for a vector destination is open coded. (reported by
4121 Brian Downing on c.l.l)
4122 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
4123 documentation string.
4124 * fixed some bugs revealed by Paul Dietz' test suite:
4125 ** COPY-ALIST now signals an error if its argument is a dotted
4127 ** Condition slots are now accessed more correctly in the presence
4128 of multiple initargs for a given slot.
4129 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
4130 exclude restarts of the same name associated with a different
4132 ** DEFCLASS of forward-referenced classes with another
4133 forward-referenced class in the superclasses list no longer
4135 ** Condition slots are now initialized once each, not multiple
4136 times. (thanks to Gerd Moellmann)
4137 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
4139 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
4140 ** NSET-EXCLUSIVE-OR does not return extra elements when its
4141 arguments contain duplicated elements.
4142 ** RESTART-CASE understands local macros.
4143 ** RESTART-CASE associates exactly its own restarts with a condition.
4144 ** ENDP in safe mode checks its argument to be of type LIST.
4145 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
4146 in question is unbound.
4147 ** Optimized MAKE-INSTANCE functions no longer cause internal
4148 assertion failures in the presence of duplicate initargs.
4149 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
4150 ** Side-effectful :DEFAULT-INITARGS have their side-effects
4151 propagated even in the ctor optimized implementation of
4153 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
4155 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
4156 * a better implementation of SXHASH on (simple) bit vectors,
4157 measured both in execution speed and in distribution of results
4158 over the positive fixnums, has been installed. Likewise, a better
4159 implementation of EQUAL for simple bit vectors is now available.
4160 * fixed CEILING optimization for a divisor of form 2^k.
4161 * fixed bug 240 (emitting extra style warnings "using the lexical
4162 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
4163 by Antonio Martinez)
4164 * fixed SXHASH, giving different results for NIL depending on type
4165 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
4166 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
4167 variable. (found by Rolf Wester)
4168 * SBCL does not ignore type declarations for special
4169 variables. (reported by rif on c.l.l 2003-03-05)
4170 * some bug fixes in contrib/sb-aclrepl/
4171 * fixed some bugs revealed by Paul Dietz' test suite:
4172 ** a bug in the CONS type specifier, whereby the CAR and CDR
4173 types got intertwined, has been fixed;
4174 ** the type system is now able to reason about the interaction
4175 between INTEGER and RATIO types more completely;
4176 ** APPEND, [N]REVERSE and NRECONC check that those their
4177 arguments, which must be proper lists, are really so;
4178 ** An array specialized to be unable to hold elements has been
4179 implemented, as required -- yes, really -- by ANSI;
4180 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4181 on malformed property lists;
4183 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4184 * incompatible packaging change: in line with Unix convention,
4185 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4186 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4187 * REQUIRE and PROVIDE are now optionally capable of doing something
4188 useful. See the documentation string for REQUIRE.
4189 * infrastructure for a managed SBCL contrib system: contributed
4190 modules in this release include:
4191 ** the ASDF system definition facility;
4192 ** an interface to the BSD Sockets API;
4193 ** an ACL-like convenience interface to the repl;
4194 (thanks to Kevin Rosenberg)
4195 ** an implementation of ROTATE-BYTE, with efficient implementation
4197 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4198 gives a closer approximation to the right answer than previously.
4199 (thanks to Raymond Toy)
4200 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4201 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4202 optional environment arguments, as required by ANSI.
4203 * fixed bugs in other functions taking environment objects, allowing
4204 calls with an explicit NIL environment argument to be compiled
4206 * fixed bug 228: primary return values from
4207 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4208 COMPILE or FUNCTION.
4209 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4210 :TYPE will no longer signal errors on innocuous objects.
4211 * fixed bug 231b: SETQ is better at respecting type declarations in
4212 the lexical environment.
4213 * fixed a bug in DEFCLASS: classes named by symbols with no or
4214 unprintable packages can now be defined.
4215 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4216 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4217 * fixed compiler failure related to checking types of functions.
4218 (reported by Robert E. Brown)
4219 * the compiler is now much more consistent in its error-checking
4220 treatment of bounding index arguments to sequence functions: in
4221 (SAFETY 3) code, errors will be signalled in almost all cases if
4222 invalid sequence bounding indices are passed to functions defined
4223 by ANSI to operate on sequences.
4224 * fixed a bug in the build procedure: documentation of SBCL-specific
4225 packages is now preserved and available in the final Lisp image.
4226 * lifted FDEFINITION lookup out of loops in the implementation of
4227 many list operations. (thanks to Robert E. Brown)
4228 * fixed a bug in the reader: the #n# reader macro now works for
4229 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4230 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4231 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4232 E. Brown and Tony Martinez respectively)
4233 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4234 SLOT-BOUNDP: in particular, it is now able to optimize them much
4235 better, and is now not vulnerable to having packages renamed.
4236 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4237 always called when a slot is not present in an instance. (thanks
4239 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4240 longer signal an error on unreading a character following EOF on
4241 the previous constituent stream. (thanks to Tony Martinez)
4242 * fixed some bugs revealed by Paul Dietz' test suite:
4243 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4244 not just nonnegative fixnums;
4245 ** the logical bit-array operators such as BIT-AND now accept an
4246 explicit NIL for their "opt-arg" argument (to indicate a
4247 freshly-consed result bit-array);
4248 ** ELT now signals an error on an invalid sequence index in safe
4250 ** the type system is now cleverer about negations of numeric
4251 types, and consequently understands the BIGNUM and RATIO types
4253 ** the type system is now cleverer about the interaction between
4254 INTEGER and RATIO types: while bugs still remain, many more
4255 cases are accurately computed;
4256 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4257 if it is in the last clause;
4258 ** CONSTANTLY now correctly returns a side-effect-free function in
4260 ** DECLARE is no longer treated as a special-operator; in
4261 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4262 * incremented fasl file version number due to the change in the
4263 DEFSTRUCT-SLOT-DESCRIPTION structure.
4265 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4266 * minor incompatible change: code processed by the "interpreter" or
4267 EVAL now has a compilation optimization policy of (DEBUG 2)
4268 (changed from (DEBUG 1)) to improve debuggability of interactive
4269 development, and to allow the use of the debug RETURN command in
4271 * an experimental implementation of the RETURN command for the
4272 debugger has been included. (thanks to Frederik Kuivinen)
4273 * fixed bug 62: constraints were not propagated into a loop.
4274 * fixed bug in embedded calls of SORT (reported and investigated by
4276 * fixed some bugs revealed by Paul Dietz' test suite:
4277 ** printing and reading of arrays with some dimensions having
4278 length 0 (thanks to Gerd Moellmann);
4279 ** BOA constructor with &AUX argument without a default value does
4280 not cause a type error;
4281 ** CONSTANTP now returns true for all self-evaluating objects.
4283 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4284 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4285 accessors that are related by inheritance, as specified in the
4286 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4287 Valtteri Vuorikoski)
4288 * The compiler is now able to inline functions that were defined in
4289 a complex lexical environment (e.g. inside a MACROLET).
4290 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4292 * Support for the upcoming FreeBSD-5.0 release has been included.
4293 (thanks to Dag-Erling Smorgrav)
4294 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4295 effect when it is not in a toplevel context.
4296 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4297 stand a better chance of being correct. (thanks to Gerd
4299 * fixed bug in COERCE, which now signals an error on coercing a
4300 rational to a bounded real type which excludes the expected
4302 * The compiler is now able to derive types more accurately from the
4303 COERCE and COMPILE functions.
4304 * fixed bug 223: functional binding is considered to be constant
4305 only for symbols in the CL package.
4306 * fixed bug 231: SETQ did not check the type of a variable being set
4307 (reported by Robert E. Brown)
4308 * A new optimization for MAKE-INSTANCE has been included, fixing
4309 various bugs (including relating to :ALLOCATION :CLASS slots and
4310 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4312 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4313 ** As required by ANSI, LOOP now disallows anonymous collection
4314 clauses such as COLLECT I in conjunction with aggregate boolean
4315 clauses such as THEREIS (= I 1);
4316 ** LOOP now signals an error when any variable is reused in the
4317 same loop (including the potentially useful construct analogous
4318 to WITH A = 1 WITH A = (1+ A);
4319 ** IT is only a special loop symbol within the first clause of a
4320 conditional loop clause;
4321 ** LOOP with a typed iteration variable over a hashtable now
4322 signals a type error iff it should.
4323 * fixed some other bugs revealed by Paul Dietz' test suite:
4324 ** FILE-STREAM now names the class previously known as FD-STREAM;
4325 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4326 argument) no longer signals an error;
4327 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4328 of the slot symbol, rather than using the current package
4329 ((:CONC-NAME "") continues to intern the slot's name in the
4331 * incremented fasl file version number, because of the incompatible
4332 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4333 of the new implementation of DEFINE-COMPILER-MACRO.
4335 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4336 * Support for building SBCL for MIPS platforms running in
4337 little-endian mode has now been checked in, and basic
4338 functionality on said platforms verified.
4339 * minor incompatible change: PCL now records the pathname of a file
4340 in which methods and the like are defined, rather than its
4342 * minor incompatible change: TRUENAME now considers the truename of
4343 a file naming a directory to be the pathname with :DIRECTORY
4344 component indicating that directory.
4345 * minor incompatible change: a NAMED clause in the extended form of
4346 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4347 reason for the previous behaviour is unclear.
4348 * more systematization and improvement of CLOS and MOP conformance
4349 in PCL (thanks to Gerd Moellman and Pierre Mai):
4350 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4352 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4353 primary methods with no specializers;
4354 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4356 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4357 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4358 FORWARD-REFERENCED-CLASSes; error reporting on
4359 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4361 ** SXHASH on CLOS instances now uses a slot internal to the
4362 instance to return different numbers on distinct instances,
4363 while preserving the same return value through invocations of
4365 ** DEFMETHOD signals errors when methods with longer incongruent
4366 lambda lists are added to generic functions;
4367 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4368 CLASS, as specified in AMOP;
4369 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4370 on the order returned by the primary method for classes of
4371 class STANDARD-CLASS;
4372 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4373 * fixed some bugs shown by Paul Dietz' test suite:
4374 ** DOLIST puts its body in TAGBODY;
4375 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4377 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4378 value producing form;
4379 ** if more variables are given to PROGV than values, extra
4380 variables are bound and made to have no value;
4381 ** NSUBSTITUTE on list arguments gets the right answer with
4383 ** ELT signals an error of type TYPE-ERROR when the index argument
4384 is not a valid sequence index;
4385 ** LOOP signals (at macroexpansion time) an error of type
4386 PROGRAM-ERROR when duplicate variable names are found;
4387 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4388 ** FUNCALL of special-operators now cause an error of type
4390 ** PSETQ now works as required in the presence of side-effecting
4391 symbol-macro places;
4392 ** NCONC accepts any object as its last argument;
4393 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4395 ** loop-for-as-package does not require a package to be explicitely
4397 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4399 * fixed bug 166: compiler preserves "there is a way to go"
4400 invariant when deleting code.
4401 * fixed bug 172: macro lambda lists with required arguments after
4402 &REST arguments now cause an error to be signalled. (thanks to
4404 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4405 bug. (thanks to Matthew Danish)
4406 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4408 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4409 arguments are lexically rebound. (thanks to Gerd Moellmann and
4411 * fixed bug 194: error messages are now more informative when there
4412 is no primary method applicable in a call to a generic
4413 function. (thanks to Gerd Moellmann)
4414 * fixed bug in command line argument checking (thanks to Julian
4416 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4417 (thanks to Matthew Danish)
4418 * incremented fasl file version number, because of the
4419 SXHASH-related changes in the layout of CLOS data structures
4421 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4422 * minor incompatible change: The runtime (the Unix executable named
4423 "sbcl") is now much pickier about the .core files it will load.
4424 Essentially it now requires .core files to descend from the same
4425 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4426 as the runtime does. (The intent is to prevent the crashes which
4427 can occur, and which can even be reported as mysterious failures,
4428 when people patch the sources or change the build parameters
4429 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4430 sbcl and .core files.)
4431 * fixed bug: VALUES-LIST is no longer optimized away.
4432 * fixed bug 142: The FFI conversion of C string values to Lisp
4433 string values no longer conses excessively. (thanks to Nathan
4434 Froyd porting Raymond Toy's fix to CMU CL)
4435 * began to systematize and improve MOP conformance in PCL (thanks to
4436 Nathan Froyd, Gerd Moellman and Pierre Mai):
4437 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4439 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4440 ** FINALIZE-INHERITANCE is now called on class finalization;
4441 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4442 argument precedence order.
4443 * fixed bug 202: The compiler no longer fails on functions whose
4444 derived types contradict their declared type.
4445 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4446 so it can be non-toplevel.
4447 * The fasl file version number has changed (because of the new
4448 implementation of DEFMACRO).
4449 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4450 safe code, that any length requirement by their type-specifier
4451 argument is valid. The exceptions to this are described in bug
4453 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4454 functions now check the types of their inputs as required by ANSI.
4455 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4456 introduced symbol is DECLAREd to be SPECIAL.
4457 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4458 * fixed bug: :COUNT argument to sequence functions may be negative
4459 * fixed bug: body of DO-SYMBOLS may contain declarations
4460 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4461 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4462 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4463 symbol macro only once
4464 * fixed printing of call frame when argument list is unavailable
4465 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4466 * compiler no longer signals WARNING on unknown keyword
4469 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4470 * A beta-quality port to the mips architecture running Linux,
4471 based on the old CMUCL backend, has been made. It has been tested
4472 on a big-endian kernel, and works sufficiently well to be able to
4473 rebuild itself; it has not been tested in little-endian mode.
4474 * fixed an inconsistency between gencgc.c and purify.c which made
4475 dumping/loading .core files unreliable
4476 * fixed bug 120a: The compiler now deals correctly with IFs where
4477 the consequent is the same as the alternative, instead of
4478 misderiving the return type. (thanks to Alexey Dejneka)
4479 * fixed bug 113: Logical pathnames are now dumpable (the logical
4480 host is resolved at load-time, throwing an error if it is not
4482 * fixed bug 174: FORMAT's error message is slightly clearer when a
4483 non-printing character is used in a format directive.
4484 * fixed several bugs in compiler checking of type declarations, i.e.
4485 violations of the Python "declarations are assertions" principle
4486 (thanks to Alexey Dejneka)
4487 * fixed several bugs in PCL's error checking (thanks to Gerd
4489 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4491 * fixed bug in compilation of functions as first class values
4492 (thanks to Antonio Martinez-Shotton)
4493 * The compiler's handling TYPE-ERRORs which it can prove will
4494 inevitably happen at runtime has been cleaned up and corrected
4495 in several ways. (thanks to Alexey Dejneka)
4496 * improved argument type checking for various basic arithmetic
4497 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4498 lavished on them in the past that they can be compiled in many
4499 ways in different special cases
4500 * fixed bug 181: compiler checks validity of user supplied type
4502 * cleaned up code flushing in optimization: Function calls which
4503 should signal errors for safety purposes (e.g. which ANSI says
4504 should signal errors when their arguments are of incorrect type)
4505 are no longer optimized away.
4506 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4507 * incremented fasl file version number, because changes in the
4508 implementation of sequence functions like COERCE caused
4509 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4510 in old inline expansions) to become undefined. (Actually these
4511 changes were later undone, so we might very well be binary
4512 compatible with 0.7.7 after all, but leaving the version number
4513 incremented seemed like the simplest and most conservative
4516 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4517 * An alpha-quality port to the parisc architecture running Linux,
4518 based on the old CMUCL backend, has been made. This, even more so
4519 than the other backends, should be considered still a work in
4520 progress; known problems include that the Linux kernel in 64-bit
4521 mode does not propagate the correct sigcontext structure to
4522 userspace, and consequently SBCL on a parisc64 kernel will not
4524 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4525 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4526 suppressed.) Also now that the compiler is looking at declarations
4527 in the environment, it checks optimization declarations as well,
4528 and suppresses inlining when (> DEBUG SPEED).
4529 * More fixes have been made to treatment of floating point exception
4530 treatment and other Unix signals. In particular, floating point
4531 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4532 * The detection and handling of control stack exhaustion (infinite
4533 or very deeply nested recursion) has changed. Stack exhaustion
4534 detection is now done by write-protecting pages at the OS level
4535 and applies at all optimization settings; when found, a
4536 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4537 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4538 be caught using IGNORE-ERRORS.
4539 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4540 correctly when transforming recursive LABELS functions to LETs.
4541 (thanks to Alexey Dejneka)
4542 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4543 that are names of constants or global variables.
4544 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4545 alien routines with docstrings.
4546 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4547 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4549 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4550 (thanks to Eric Marsden)
4551 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4552 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4553 * Bug fix: The SPARC backend can now compile functions involving
4554 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4555 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4556 to a functions expecting a PATHNAME-DESIGNATOR.
4557 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4558 lambda lists. (thanks to Alexey Dejneka)
4559 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4560 functions so that EXPORTing the name of the class doesn't cause
4561 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4562 to Antonio Martinez for reporting this)
4563 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4564 constant-folded without failing an assertion. (thanks to Einar
4565 Floystad Dorum for reporting this)
4566 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4567 types are now dealt with more correctly by the compiler.
4568 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4569 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4570 behaviour analogous to RENAME-FILE. This puts its behaviour more
4571 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4573 * The fasl file version number has changed again. (because of the
4574 bug fix involving the names of PCL MAKE-INSTANCE functions)
4576 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4577 * bug fix: Floating point exceptions are treated much more
4578 consistently on the x86/Linux and PPC/Linux platforms.
4579 * Array initialization with :INITIAL-ELEMENT is now much faster for
4580 cases when the compiler cannot open code the array creation, but
4581 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4582 array accesses have also seen a speed increase.
4583 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4584 specified. (This was at the root of some bad interactions between
4585 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4587 * bug fix: Internal error arguments for undefined functions are now
4588 computed correctly on the PPC/Linux platform.
4589 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4590 Raymond Toy's patch for CMU CL)
4591 * Support for the Solaris 9 operating environment has been included
4592 (thanks to Daniel Merritt)
4593 * A very ugly but hopefully complete draft of the missing FFI chapter
4594 of the manual has been created by reformatting the corresponding
4595 CMU CL manual chapter into (currently very ugly and incoherent)
4596 DocBook and bringing it up to date for SBCL behavior. Thus, the
4597 manual is now essentially complete, at least by my extreme
4598 once-and-only-once standards, whereby it's acceptable to refer to
4599 the doc strings of SB-EXT functions as the primary documentation.
4600 * The fasl file version number has changed again, due to cleanup of
4601 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4602 is no longer a static symbol.)
4604 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4605 * SBCL now builds with OpenMCL (version 0.12) as the
4606 cross-compilation host; also, more progress has been made toward
4607 bootstrapping under CLISP.
4608 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4610 * bug 158 fixed: The compiler can now deal with integer loop
4611 increments different from 1; fixing this turned out also to fix
4613 * bug 169 fixed: no more bogus warnings about using lexical bindings
4614 despite the presence of perfectly good SPECIAL declarations (thanks
4615 to David Lichteblau)
4616 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4617 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4618 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4620 * bug 180 fixed: Method combination specifications no longer ignore
4621 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4622 * bug fix: Structure type predicate functions now check their argument
4623 count as they should.
4624 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4625 correctly. (thanks to Pierre Mai)
4626 * minor incompatible change: The --noprogrammer option is deprecated
4627 in favor of the new --disable-debugger option, which is very similar.
4628 (The major difference is that it takes effect at a slightly different
4629 time at startup, causing handling of errors in --sysinit and
4630 --userinit files will be affected differently.) The
4631 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4632 been added to allow this functionality to be controlled from ordinary
4633 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4634 maintainers, who might want to run non-interactive scripts to
4635 build SBCL cores which will later be used interactively.)
4636 * minor incompatible change: The LOAD function no longer, when given
4637 a wild pathname to load, loads all files matching that pathname.
4638 Instead, an error of type FILE-ERROR is signalled.
4640 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4641 * bug 147 fixed: The compiler preserves its block link/count
4642 invariants more correctly now so that it doesn't crash. (thanks
4644 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4646 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4647 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4648 no-value-for-this-character as NIL (as specified by ANSI).
4649 * HOST-NAMESTRING on physical pathnames now returns a string that is
4650 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4651 (thanks to Christophe Rhodes)
4652 * The Alpha port handles icache flushing more correctly. (thanks to
4654 * More progress has been made toward bootstrapping under CLISP. (thanks
4655 to Christophe Rhodes)
4656 * The fasl file format has changed again, because dynamic loading
4657 on OpenBSD (which has non-ELF object files) motivated some cleanups
4658 in the way that foreign symbols are transformed and passed around.
4659 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4660 is no longer treated as whitespace by the reader, but instead as
4661 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4662 |A
\7fB|, instead of A as it used to.
4664 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4665 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4666 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4667 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4668 this as of 0.7.1.45, but was left out of the previous news section
4669 (thanks to Dan Barlow)
4670 * SBCL now runs on the Solaris operating system on SPARC architectures
4671 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4672 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4673 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4674 * SPARC backend cleanups, allowing builds of cores optimized for V8
4675 and V9 SPARCS, and also emission of code targeted to a particular
4676 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4678 * SBCL is closer to bootstrapping under CLISP, thanks to various
4679 fixes by Christophe Rhodes.
4680 * The fasl file format has changed again, to allow the compiler's
4681 INFO database to support symbol macros.
4682 * The user manual (in doc/) is formatted into HTML more nicely.
4683 (thanks to coreythomas)
4684 * The system is smarter about SUBTYPEP relationships, especially
4685 those involving NOT types (including types such as ATOM which are
4686 represented internally using NOT types). Thus SUBTYPEP is less
4687 likely to return (VALUES NIL NIL) in general, and in particular
4688 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4690 * The fasl file format has changed again, because the internal
4691 representation of types now includes a new slot to support the new
4692 SUBTYPEP-of-NOT-types logic.
4693 * (not a change in the main branch of SBCL, but a related prototype
4694 which can hopefully be merged into the main branch of SBCL in the
4695 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4696 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4697 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4698 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4699 dispatch macro character combinations. (thanks to Alexey Dejneka)
4700 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4701 that superficially look like logical namestrings correctly.
4702 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4704 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4705 default supplied-p) for &optional and &key arguments. (thanks to
4708 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4709 * incompatible change: The compiler is now less aggressive about
4710 tail call optimization, doing it only when (> SPACE DEBUG) or
4711 (> SPEED DEBUG). (This is an incompatible change because there are
4712 programs which relied on the old CMU-CL-style behavior to optimize
4713 away their unbounded recursion which will now die of stack overflow.)
4714 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4715 for non-GENCGC systems has been increased to 20M (since that
4716 seems much closer to the likely performance optimum for modern
4717 systems than the old 4M value was)
4718 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4719 and *DEBUG-PRINT-LEVEL*
4720 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4721 of CMU CL's support for SPARC, and various endianness and other
4722 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4723 * new syntactic sugar for the Unix command line: --load foo.bar is now
4724 an alternate notation for --eval '(load "foo.bar")'.
4726 ** The system now detects stack overflow and handles it gracefully,
4727 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4728 optimization settings. (This is a good thing in general, and
4729 its introduction in this version should be particularly timely
4730 for anyone whose code fails because of suppression of tail
4732 ** The system now hunts for the C variable "environ" in a more
4733 devious way, to avoid segfaults when the C library version differs
4734 between compile time and run time. (thanks to Christophe Rhodes)
4735 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4736 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4737 ** The compiler no longer issues bogus style warnings for undefined
4738 classes in the same source file as the DEFCLASSes which defined
4739 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4741 ** fixes in CONDITION class precedence list for undefined function
4742 errors (thanks to Alexei Dejneka)
4743 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4744 correctly. (thanks to Dan Barlow)
4745 ** portability fixes aiming at bootstrapping under CLISP (thanks
4746 to Dave McDonald and Christophe Rhodes)
4747 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4748 ** fixes in type translation and and type inference (thanks to
4750 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4751 ** various fixes in the new ports (thanks to Dan Barlow)
4752 * several changes related to debugging:
4753 ** suppression of tail recursion, as noted above
4754 ** stack overflow detection, as noted above
4755 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4756 is now the default. (For some time encapsulation has been more
4757 reliable than the breakpoint-based :ENCAPSULATE NIL
4758 implementation, at least on X86 systems; and I just noticed that
4759 encapsulation also seems closer to the spirit of the ANSI
4762 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4764 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4765 up properly again. (There was a packaging bug in 0.7.0 which
4766 left their definitions in SB-SYS::LOAD-FOREIGN and
4767 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4768 vital for most things which interface to C-level interfaces,
4769 like extensions working with sockets or databases or
4770 Perl-compatible regexes or whatever, and the need to fix
4771 this bug is the main reason that 0.7.1 was released so
4773 ** DEFGENERIC is now choosier about the methods it redefines, so that
4774 reLOADing a previously-LOADed file containing DEFGENERICs does
4775 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4776 cycle now works as it should. (thanks to Alexey Dejneka)
4777 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4778 fixed by Christophe Rhodes. (He actually submitted this patch
4779 months ago, and I delayed until after 0.7.0.)
4780 ** Bug 111 (internal compiler confusion about runtime checks on
4781 FUNCTION types) was fixed by Alexey Dejneka.
4782 * Some internal cleanups (getting rid of variables which aren't
4783 needed now that the byte interpreter is gone) caused the fasl
4784 file format number to change again.
4786 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4787 * major incompatible change: The default fasl file extension, i.e. the
4788 default extension for files produced by COMPILE-FILE, has changed
4789 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4791 ** There are many changes in the implementation of the compiler.
4792 SBCL is now essentially a compiler-only implementation of ANSI
4793 Common Lisp. EVAL still "interprets" a few special cases, but
4794 almost all the interesting cases are handled by creating
4795 a LAMBDA expression, calling COMPILE on it, then calling
4796 FUNCALL on the result.
4797 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4798 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4799 Since the code is newer, there might still be some new bugs
4800 (though not as many as before Martin Atzmueller's fixes:-). But
4801 the new code is substantially simpler and clearer, and hopefully
4802 any remaining bugs will be simpler, less fundamental, and more
4803 fixable then the bugs in the old code.
4804 ** The revised compiler is still a little unsteady on its feet.
4806 *** The debugging information it produces (particularly the names
4807 of FUNCTION objects) is sometimes much less useful than what
4808 the old compiler produced.
4809 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4810 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4811 within a MACROLET) has been temporarily weakened.
4812 ** There are new compiler optimizations for various functions:
4813 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4814 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4815 *** the math functions TRUNCATE, FLOOR, and CEILING
4816 *** the function-of-all-trades COERCE
4817 Mostly these should be transparent, but there's one
4818 potentially-annoying problem (bug 117): when the compiler
4819 inline-expands a function and does type analysis on the result,
4820 it can create control paths which have type mismatches, and
4821 when it can't prove that those control paths aren't taken,
4822 it will issue WARNINGs about the type mismatches. This is
4823 a particular problem in practice for the new sequence functions.
4824 It's not clear how this should be fixed, and for now, a
4825 workaround is given in the entry for 117 in the BUGS file.
4826 ** (Because of the interaction between the two previous items --
4827 occasional inlining problems and new inline expansions -- some
4828 of the new sequence function optimizations won't really kick in
4829 completely until debugging information, and then inlining, are
4830 straightened out in some future version.)
4831 * minor incompatible changes:
4832 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4833 DIRECTORY no longer implicitly promotes NIL slots of its
4834 pathname argument to :WILD. In particular, when you ask for the
4835 contents of a directory (which you used to be able to do without
4836 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4837 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4838 ** changes in behavior that ANSI explicitly defines to be
4839 implementation dependent:
4840 *** The new compiler-only implementation still conforms with ANSI,
4841 but acts a little different than before. Besides the obvious
4842 changes in performance tradeoffs (that the cost per form passed
4843 to EVAL has gone up, and the cost per form executed by EVAL
4844 has gone down), the behavior of the system changes a little
4845 because there are no longer any interpreted function objects.
4846 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4847 e.g. doing COMPILE on the output of interactive DEFUN is
4849 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4851 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4852 has changed from (:CL) to NIL.
4853 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4854 e.g. #\Newline and #\Tab, don't have names specified in the
4855 ANSI Common Lisp standard, is now based on their ASCII symbolic
4856 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4857 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4858 are no longer used for output.
4859 ** changes in internal implementation constants:
4860 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4861 4 million. (If your application spends a lot of time GCing and
4862 you have a lot of RAM, you might want to experiment with
4863 increasing it even more.)
4864 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4865 However, almost all old code should still continue to work without
4866 immediate update, as SB-C-CALL is now a (deprecated) nickname
4868 ** Old operator names in the style DEF-FOO are now deprecated in
4869 favor of new corresponding names DEFINE-FOO, for consistency with
4870 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4871 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4872 internal symbols, but a few supported extensions like
4873 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4874 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4875 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4876 etc. as you get deeper into recursive calls to the debugger
4877 command loop, instead of the old "5]", "5]]", "5]]]"
4878 sequence. (I was motivated to do this when squabbles between
4879 ILISP and SBCL left me very deeply nested in the debugger. In the
4880 short term, this change will probably provoke more ILISP/SBCL
4881 squabbles, but hopefully it will be an improvement in the long run.)
4882 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4883 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4884 function, since declaiming return types involving aliens is
4885 (1) annoyingly messy to do by hand and (2) vital to efficient
4886 compilation of code which calls such functions.
4887 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4888 longer reexported by the SB-EXT package. They're solely useful
4889 for alien code, so it seems more logical that you should get
4890 them from the SB-ALIEN package, not in SB-EXT.
4891 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4892 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4893 features. Instead, the code that they used to control is always
4894 built into the system.
4895 * many other bug fixes
4896 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4897 advantage of the new EVAL-WHEN stuff and to clean them up in
4898 general, and they are now more ANSI-compliant in a number of
4899 ways. Martin Atzmueller is responsible for a lot of this.
4900 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4902 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4903 *** correct ERROR type for various file operations
4904 *** some fixes for Lisp streams
4905 *** DEFMETHOD syntax checking
4906 *** changing old weird representation of debug information as
4907 strings (which, among their other deficiencies, don't transform
4908 correctly when you rename packages, and don't change their
4909 print representation when you change things like *PACKAGE*
4910 and *PRINT-LENGTH*) to symbols and lists of symbols
4911 He also made several improvements and fixed several bugs in DESCRIBE.
4912 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4914 *** misbehavior of WRITE-STRING/WRITE-LINE
4915 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4916 and several other LOOP problems as well
4917 *** DIRECTORY when similar filenames are present
4918 *** DEFGENERIC with :METHOD options
4919 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4920 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4921 *** argument ordering in FIND with :TEST option
4922 *** mishandled package designator argument in APROPOS-LIST
4923 *** various problems in the backquote readmacro
4924 *** a bug in APROPOS
4925 *** probably some others that I'm not describing very well here,
4926 since the CVS log documents them by reference to sbcl-devel
4927 messages, and the SourceForge archives aren't working well.:-(
4928 ** Dan Barlow improved the Alpha port (and is making progress on the
4929 PPC port, for those of you who think different).
4930 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4931 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4932 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4933 continues to work on the SPARC port (for those of you in a position
4934 to look down upon our little PC-compatible boxes from a great height).
4935 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4936 and uses that copy, rather than the current dynamic value, when
4937 it's trying to decide whether to truncate output. Thus e.g.
4938 (let ((*print-lines* 50))
4939 (pprint-logical-block (stream nil)
4941 (let ((*print-lines* 8))
4942 (print (aref possiblybigthings i) stream)))))
4943 should now truncate the logical block only at 50 lines, instead of
4944 often truncating it at 8 lines, as it did before.
4945 * The doc/cmucl/ directory, containing old CMU CL documentation
4946 from the time of the fork, is no longer part of the base system.
4947 SourceForge has shut down its anonymous FTP service, and with it
4948 my original plan for distributing the old CMU CL documentation
4949 there. For now, if you need these files you can download an old
4950 SBCL source release and extract them from it.
4951 * The fasl file version number changed again, for dozens of reasons,
4952 some of which are apparent above.
4954 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4955 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4956 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4957 patch, so that DISASSEMBLE output is much nicer.
4958 * The code in the SB-PROFILE package now seems reasonably stable.
4959 I still haven't decided what the final interface should look like
4960 (I'd like PROFILE to interact cleanly with TRACE, since both
4961 facilities use function encapsulation) but if you have a need
4962 for profiling now, you can probably use it successfully with
4963 the current CMU-CL-style interface.
4964 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4965 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4966 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4967 link now returns the dangling link itself, and for similar
4968 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4969 link itself. (In these cases the old code signalled an error and
4970 looped endlessly, respectively.) Thus, DIRECTORY now works even
4971 in the presence of dangling and cyclic symbolic links.
4972 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4973 is now a supported extension again, since the consensus on
4974 sbcl-devel was that it can be useful for ordinary development
4975 work, not just for debugging SBCL itself.
4976 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4977 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4978 that currently-defined functions might be redefined later with
4979 different return types.
4980 * Hash tables can be printed readably, as inspired by CMU CL code
4981 of Eric Marsden and SBCL code of Martin Atzmueller.
4982 * better error handling in CLOS method combination, thanks to
4983 Martin Atzmueller porting Pierre Mai's CMU CL patches
4984 * more overflow fixes for >16Mbyte I/O buffers
4985 * A bug in READ has been fixed, so that now a single Ctrl-D
4986 character suffices to cause end-of-file on character streams.
4987 In particular, now you only need one Ctrl-D at the command
4988 line (not two) to exit SBCL.
4989 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4991 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4992 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4993 does the right thing.
4994 * The implementation of some type tests, especially for CONDITION
4995 types, is now tidier and maybe faster, due to CMU CL code
4996 originally by Douglas Crosher, ported by Martin Atzmueller.
4997 * Some math functions have been fixed, and there are new
4998 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4999 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
5000 * (There are also some new optimizers in contrib/*-extras.lisp. Those
5001 aren't built into sbcl-0.6.13, but are a sneak preview of what's
5002 likely to be built into sbcl-0.7.0.)
5003 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
5004 CL, ported to SBCL by Martin Atzmueller)
5005 * DESCRIBE now gives more information in some cases. (Pierre Mai's
5006 patch to CMU CL, ported to SBCL by Martin Atzmueller)
5007 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
5008 * There's a new slam.sh hack to shorten the edit/compile/debug
5009 cycle for low-level changes to SBCL itself, and a new
5010 :SB-AFTER-XC-CORE target feature to control the generation of
5011 the after-xc.core file needed by slam.sh.
5012 * minor incompatible change: The ENTRY-POINTS &KEY argument to
5013 COMPILE-FILE is no longer supported, so that now every function
5014 gets an entry point, so that block compilation looks a little
5015 more like the plain vanilla ANSI section 3.2.2.3 scheme.
5016 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
5017 returns the number of bytes consed since the system started,
5018 rather than the number consed since the first time the function
5019 was called. (The new definition parallels ANSI functions like
5020 CL:GET-INTERNAL-RUN-TIME.)
5021 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
5022 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
5023 supported. Now DIRECTORY always does the abstract Common-Lisp-y
5024 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
5025 * Fasl file version numbers are now independent of the target CPU,
5026 since historically most system changes which required version
5027 number changes have affected all CPUs equally. Similarly,
5028 the byte fasl file version is now equal to the ordinary
5031 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
5032 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
5033 is no longer recognized. I apologize for this, because it was
5034 listed in SB-EXT as a supported extension, but I found that
5035 its existing behavior was poorly specified, as well as incorrectly
5036 specified, and it looked like too much of a mess to straighten it
5037 out. I have enough on my hands trying to get ANSI stuff to work..
5038 * many patches ported from CMU CL by Martin Atzmueller, with
5039 half a dozen bug fixes in pretty-printing and the debugger, and
5040 half a dozen others elsewhere
5041 * fixed bug 13: Floating point infinities are now supported again.
5042 They might still be a little bit flaky, but thanks to bug reports
5043 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
5044 as flaky as they were.
5045 * The --noprogrammer command line option is now supported. (Its
5046 behavior is slightly different in detail from what the old man
5047 page claimed it would do, but it's still appropriate under the
5048 same circumstances that the man page talks about.)
5049 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
5050 are now supported, and enabled by default. Thus, the compiler can
5051 handle many floating point and complex operations much less
5052 inefficiently. (Thus e.g. you can implement a complex FFT
5054 * The compiler now detects type mismatches between DECLAIM FTYPE
5055 and DEFUN better, and implements CHECK-TYPE more correctly, and
5056 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
5057 thanks to patches from Martin Atzmueller.
5058 * various fixes to make the cross-compiler more portable to
5059 ANSI-conforming-but-different cross-compilation hosts (notably
5060 Lispworks for Windows, following bug reports from Arthur Lemmens)
5061 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
5062 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
5063 fixed thanks to Pierre Mai's CMU CL patches.
5064 * fixes to keep the system from overflowing internal counters when
5065 it tries to use i/o buffers larger than 16M bytes
5066 * fixed bug 45a: Various internal functions required to support
5067 complex special functions have been merged from CMU CL sources.
5068 (When I was first setting up SBCL, I misunderstood a compile-time
5069 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
5070 * improved support for type intersection and union, fixing bug 12
5071 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
5072 more obscure bugs as well
5073 * some steps toward byte-compiling non-performance-critical
5074 parts of the system, courtesy of patches from Martin Atzmueller
5075 * Christophe Rhodes has made some debian packages of sbcl at
5076 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
5077 From his sbcl-devel e-mail of 2001-04-08 they're not completely
5078 stable, but are nonetheless usable. When he's ready, I'd be happy
5079 to add them to the SourceForge "File Releases" section. (And if
5080 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
5081 * new fasl file format version number (because of changes in
5082 internal representation of (OR ..) types to accommodate the new
5083 support for (AND ..) types, among other things)
5085 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
5086 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
5088 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
5090 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
5091 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
5092 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
5093 * DESCRIBE now works on structure objects again.
5094 * Most function call argument type mismatches are now handled as
5095 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
5096 know whether the function will be redefined before the call is
5097 executed. (The compiler could flag local calls with full WARNINGs,
5098 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
5099 it doesn't keep track of enough information to know whether calls
5100 are local in this sense.)
5101 * Compiler output is now more verbose, with messages truncated
5102 later than before. (There should be some supported way for users
5103 to override the default verbosity, but I haven't decided how to
5104 provide it yet, so this behavior is still controlled by the internal
5105 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
5106 src/compiler/ir1util.lisp.)
5107 * Fasl file format version numbers have increased again, because
5108 support for the Gray streams extension changes the layout of the
5109 system's STREAM objects.
5110 * The Gray subclassable streams extension now works, thanks to a
5111 patch from Martin Atzmueller.
5112 * The full LOAD-FOREIGN extension (not just the primitive
5113 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
5114 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
5115 but like most other programs, it defaults to copying the Unix
5116 environment from the original process instead of starting the
5117 new process in an empty environment.
5118 * Extensions which manipulate the Unix environment now support
5119 an :ENVIRONMENT keyword option which doesn't smash case or
5120 do other bad things. The CMU-CL-style :ENV option is retained
5121 for porting convenience.
5122 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
5123 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
5125 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
5127 * A patch from Martin Atzmueller seems to have solved the SIGINT
5128 problem, and as far as we know, signal-handling now works cleanly.
5129 (If you find any new bugs, please report them!)
5130 * The system no longer defaults Lisp source file names to types
5131 ".l", ".cl", or ".lsp", but only to ".lisp".
5132 * The compiler no longer uses special default file extensions for
5133 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
5134 seems to expect a single default extension for all compiled code,
5135 and there's no compelling reason to try to stretch the standard
5136 to allow two different extensions.) Instead, byte-compiled files
5137 default to the same extension as native-compiled files.
5138 * Fasl file format version numbers have increased again, because
5139 a rearrangement of internal implementation packages made some
5140 dumped symbols in old fasl files unreadable in new cores.
5141 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
5142 many fewer weird special cases.
5143 * Bug #17 (differing COMPILE-FILE behavior between logical and
5144 physical pathnames) has been fixed, and some related misbehavior too,
5145 thanks to a patch from Martin Atzmueller.
5146 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
5147 by Tim Moore, ported to SBCL by Martin Atzmueller.
5148 * Martin Atzmueller fixed several filesystem-related problems,
5149 including bug #36, in part by porting CMU CL patches, which were
5150 written in part by Paul Werkowski.
5151 * More compiler warnings in src/runtime/ are gone, thanks to
5152 more patches from Martin Atzmueller.
5153 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
5156 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
5158 * DESCRIBE now works on CONDITION objects.
5159 * The debugger now handles errors which arise when trying to print
5160 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
5162 * The build system now uses an additional file, customize-target-features.lisp,
5163 to allow local modifications to the target *FEATURES* list. (The point of
5164 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
5165 debugging features enabled, without having to worry about propagating it
5166 into everyone's system when I do a "cvs update".) When no
5167 customize-target-features.lisp file exists, the target *FEATURES* list
5168 should be constructed the same way as before.
5169 * fixed bugs in DEFCONSTANT ANSI-compatibility:
5170 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
5171 warn about behavior which is undefined under the ANSI spec. Note: This
5172 is specified by ANSI, but it's not very popular with programmers.
5173 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
5174 macro in the SBCL sources for an example of a workaround which you
5175 might use to make portable ANSI-standard code which does what you want.
5176 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5177 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5178 when it's used as a non-toplevel form. (This is required in order
5179 to implement the DEFCONSTANT-EQX macro.)
5180 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5181 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5182 funny things, due to bugs in the implementation of EVAL-WHEN
5183 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5184 non-toplevel DEFCONSTANTs are uncommon.)
5185 * The core file version number and fasl file version number have been
5186 incremented, because the old noncompliant DEFCONSTANT behavior involved
5187 calling functions which no longer exist, and because I also took the
5188 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5189 * fixed bug 1 (error handling before read-eval-print loop starts), and
5190 redid debugger restarts and related debugger commands somewhat while
5192 ** The QUIT debugger command is gone, since it did something
5193 rather different than the SB-EXT:QUIT command, and since it never
5194 worked properly outside the main toplevel read/eval/print loop.
5195 Invoking the new TOPLEVEL restart provides the same functionality.
5196 ** The GO debugger command is also gone, since you can just invoke
5197 the CONTINUE restart directly instead.
5198 ** The TOP debugger command is also gone, since it's redundant with the
5199 FRAME 0 command, and since it interfered with abbreviations for the
5201 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5203 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5204 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5205 some porting work by Martin Atzmueller)
5206 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5207 with a boatload of patches which he ported to SBCL. Now that those
5209 ** The system tries to make sure that its low-priority messages
5210 are prefixed by semicolons, to help people who like to use
5211 syntax highlighting in their ILISP buffer. (This patch
5212 was originally due to Raymond Toy.)
5213 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5214 patches originally written by Raymond Toy.
5215 ** The compiler understands coercion between single-value and
5216 multiple-VALUES type expressions better, getting rid of some very
5217 weird behavior, thanks to patches originally by Robert MacLachlan
5218 and Douglas Crosher.
5219 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5220 lambda lists, thanks to a patch originally by Pierre Mai.
5221 ** The system no longer bogusly warns about "abbreviated type
5223 ** The compiler gets less confused by inlining and RETURN-FROM,
5224 thanks to some patches originally by Tim Moore.
5225 ** The system no longer hangs when dumping circular lists to fasl
5226 files, thanks to a patch originally from Douglas Crosher.
5227 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5228 undefined function error.
5229 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5230 in structure slot initforms) for now, documented workaround instead:-|
5231 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5232 * fixed bug 5: added stubs for various Gray stream functions called
5233 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5234 installed, at least appropriate type errors are generated
5235 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5236 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5237 consistently in DEFMETHOD forms.
5238 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5239 it doesn't seem to affect SBCL after all
5240 * The C runtime system now builds with better optimization and many
5241 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5243 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5245 * The system is now under CVS at SourceForge (instead of the
5246 CVS repository on my home machine).
5247 * The new signal handling code has been tweaked to treat register
5248 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5249 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5250 downstream, e.g. in debug-int.lisp, has implicit dependencies
5251 on the unsignedness of integer representation of machine words,
5252 and that caused the system to bomb out with infinite regress
5253 when trying to recover from type errors involving signed values,
5254 e.g. (BUTLAST '(1 2 3) -1).)
5255 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5256 (This was one of the bugs Peter Van Eynde reported back in July.)
5257 * The system now uses code inspired by Colin Walters' O(N)
5258 implementation of MAP (from the cmucl-imp@cons.org mailing
5259 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5260 inline the MAP operation, and there is more than one
5261 sequence argument to the MAP call (so that it can't just
5262 do ETYPECASE once and for all based on the type of the
5263 single sequence argument). (The old non-inline implementation
5264 of the general M-argument sequence-of-length-N case required
5265 O(M*N*N) time when any of the sequence arguments were LISTs.)
5266 * The QUIT :UNIX-CODE keyword argument has been renamed to
5267 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5268 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5270 * Raymond Wiker's patches to port dynamic loading from Linux to
5271 FreeBSD have been added.
5272 * The BUGS file is now more nearly up to date, thanks in large part
5273 to Martin Atzmueller's review of it.
5274 * The debugger now flushes standard output streams before it begins
5275 its output ("debugger invoked" and so forth).
5276 * The core version number and fasl file version number have both
5277 been incremented, because of incompatible changes in the layout
5279 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5280 used to not be called for a saved Lisp image.)
5281 * Martin Atzmueller's version of a patch to fix a compiler crash,
5282 as posted on sbcl-devel 13 September 2000, has been installed.
5283 * Instead of installing Martin Atzmueller's patch for the
5284 compiler transform for SUBSEQ, I deleted the compiler transform,
5285 and transforms for some similar consing operations.
5286 * A bug in signal handling which kept TRACE from working on OpenBSD
5288 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5289 away by constant folding
5290 * The system now defines its address space constants in one place
5291 (in the Lisp sources), and propagates them automatically elsewhere
5292 (through GENESIS and the sbcl.h file). Therefore, patching the
5293 address map is less unnecessarily tedious and error-prone. The
5294 Lisp names of address space constants have also been systematized.
5295 * CVS tags like dollar-Header-dollar have been removed from
5296 the sources, because they have never saved me trouble and
5297 they've been source of trouble working with patches and other
5298 diff-related operations.
5299 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5300 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5302 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5304 * The system has been ported to OpenBSD.
5305 * The system now compiles with a simple "sh make.sh" on the systems
5306 that it's supported on. I.e., now you no longer need to tweak
5307 text in the target-features.lisp-expr and symlinks in src/runtime/
5308 by hand, the make.sh takes care of it for you.
5309 * The system is no longer so grossly inefficient when compiling code
5310 involving vectors implemented as general (not simple) vectors (VECTOR T),
5311 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5312 which dares to use the various sequence functions on non-simple
5313 vectors, takes less of a performance hit.
5314 * There is now a primitive type predicate VECTOR-T-P
5315 to test for the (VECTOR T) type, so that e.g.
5316 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5317 can now be compiled with some semblance of efficiency. (The old code
5318 turned the type declaration into a full call to %TYPEP at runtime!)
5319 * AREF on (VECTOR T) is still not fast, since it's still compiled
5320 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5321 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5322 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5323 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5324 without making full calls to SUBTYPEP at runtime.
5325 (Some analogous efficiency issues for non-simple vectors specialized to
5326 element types other than T, or for non-simple multidimensional arrays,
5327 have not been addressed. They could almost certainly be handled the
5328 same way if anyone is motivated to do so.)
5329 * The changes in array handling break binary compatibility, so
5330 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5331 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5332 instead of (VALUES T T).
5333 * By following the instructions that Dan Barlow posted to sbcl-devel
5334 on 2 July 2000, I was able to enable primitive dynamic object
5335 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5336 functionality is not implemented (since it calls ld to resolve
5337 library references automatically, requiring RUN-PROGRAM for its
5338 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5339 try to resolve library references) is now supported.
5340 * The system now flushes the standard output streams when it terminates,
5341 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5342 them at several other probably-convenient times, e.g. in each pass of
5343 the toplevel read-eval-print loop, and after evaluating a form given
5344 as an "--eval" command-line option. (These changes were motivated by a
5345 discussion of stream flushing issues on cmucl-imp in August 2000.)
5346 * The source transform for TYPEP of array types no longer assumes
5347 that an array whose element type is a not-yet-defined type
5348 is implemented as an array of T, but instead punts, so that the
5349 type will be interpreted at runtime.
5350 * There is now some support for cross-compiling in make.sh: each of
5351 the phases of make.sh has its own script. (This should be transparent
5352 to people doing ordinary, non-cross-compile builds.)
5353 * Since my laptop doesn't have hundreds of megabytes of memory like
5354 my desktop machine, I became more motivated to do some items on
5355 my to-do list in order to reduce the size of the system a little:
5356 ** Arrange for various needed-only-at-cold-init things to be
5357 uninterned after cold init. To support this, those things have
5358 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5359 symbols with such names are now uninterned after cold init).
5360 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5361 fasl files for cold load.
5362 ** Remove the old compiler structure pooling code (which used to
5363 be conditional on the target feature :SB-ALLOC) completely.
5364 ** Redo the representation of some data in cold init to be more compact.
5365 (I also looked into supporting byte compiled code at bootstrap time,
5366 which would probably reduce the size of the system a lot, but that
5367 looked too complicated, so I punted for now.)
5368 * The maximum signal nesting depth in the src/runtime/ support code has
5369 been reduced from 4096 to 256. (I don't know any reason for the very
5370 large old value. If the new smaller value turns out to break something,
5371 I'll probably just bump it back up.)
5372 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5374 * Many, many bugs reported by Peter Van Eynde have been added to
5375 the BUGS list; some have even been fixed.
5376 * While enabling dynamic object file loading, I tried to make the
5377 code easier to understand, renaming various functions and variables
5378 with less ambiguous names, and changing some function calling
5379 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5380 * While trying to figure out how to do the OpenBSD port, I tried to
5381 clean up some of the code in src/runtime/. In particular, I dropped
5382 support for non-POSIX signal handling, added various comments,
5383 tweaked the code to reduce the number of compilation warnings, and
5384 renamed some files to increase consistency.
5385 * To support the new automatic configuration functionality in make.sh,
5386 the source file target-features.lisp-expr has been replaced with the
5387 source file base-target-features.lisp-expr and the machine-generated
5388 file local-target-features.lisp-expr.
5389 * fixed a stupid quoting error in make.sh so that using CMU CL
5390 "lisp -batch" as cross-compilation host works again
5392 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5394 * DESCRIBE no longer tries to call itself recursively to describe
5395 bound/fbound values, so that it no longer fails on symbols which are
5396 bound to themselves (like keywords, T, and NIL).
5397 * DESCRIBE now works on generic functions.
5398 * The printer now prints less-screwed-up representations of closures
5399 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5400 * A private symbol is used instead of the :EMPTY keyword previously
5401 used to mark empty slots in hash tables. Thus
5402 (DEFVAR *HT* (MAKE-HASH-TABLE))
5403 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5404 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5405 now does what ANSI says that it should. (You can still get
5406 similar noncompliant behavior if bang on the hash table
5407 implementation with all the symbols you get back from
5408 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5409 This breaks binary compatibility, since tests for equality to
5410 :EMPTY are wired into things like the macroexpansion of
5411 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5413 * There's now a minimal placeholder implementation for CL:STEP,
5414 as required by ANSI.
5415 * An obscure bug in the interaction of the normal compiler, the byte
5416 compiler, inlining, and structure predicates has been patched
5417 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5418 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5420 * Missing ordinary arguments in a macro call are now detected even
5421 when the macro lambda list contains &KEY or &REST.
5422 * The debugger no longer complains about encountering the top of the
5423 stack when you type "FRAME 0" to explicitly instruct it to go to
5424 the top of the stack. And it now prints the frame you request even
5425 if it's the current frame (instead of saying "You are here.").
5426 * As specified by ANSI, the system now always prints keywords
5427 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5429 * The default initial SIZE of HASH-TABLEs is now smaller.
5430 * Type information from CLOS class dispatch is now propagated
5431 into DEFMETHOD bodies, so that e.g.
5432 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5434 is now basically equivalent to
5435 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5436 (DECLARE (TYPE SINGLE-FLOAT X))
5438 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5439 operation, without having to do run-time type dispatch.
5440 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5441 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5442 * Since I don't seem to be making big file reorganizations very often
5443 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5444 is overflowing my ability to conveniently back them up), I've finally
5445 checked the system into CVS. (The CVS repository is on my home system,
5446 not at SourceForge -- putting it on SourceForge might come later.)
5447 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5448 high-level GC-NOTIFY-FOO functions send their output. (There's
5449 still very little control of where low-level verbose GC functions
5450 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5451 less than it used to -- the GC-NOTIFY-FOO functions are now under
5452 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5453 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5454 in only one place in the source code, and propagates it automatically
5455 everywhere that it's needed. Thus e.g. when I bump the version from
5456 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5457 * The C source files now include boilerplate legalese and documentation
5458 at the head of each file (just as the Lisp source files already did).
5459 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5460 to his page will be replaced with a link to his new CLiki service.
5462 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5464 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5465 * The build process now looks for GNU make under the default name "gmake",
5466 instead of "make" as it used to. If GNU make is not available as "gmake"
5467 on your system, you can change this default behavior by setting the
5468 GNUMAKE environment variable.
5469 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5470 can build without error under CMU CL.
5472 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5474 * There is now a partial SBCL user manual (with some new text and some
5475 text cribbed from the CMU CL manual).
5476 * The beginnings of a profiler have been added (starting with the
5477 CMU CL profiler and simplifying and cleaning up). Eventually the
5478 main interface should be through the TRACE macro, but for now,
5479 it's still accessed through vaguely CMU-CL-style functions and macros
5480 exported from the package SB-PROFILE.
5481 * Some problems left over from porting CMU CL to the new
5482 cross-compilation bootstrap process have been cleaned up:
5483 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5484 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5485 part when getting rid of the compiler *BACKEND* stuff.)
5486 ** The value of *NULL-TYPE* was screwed up, because it was
5487 being initialized before the type system knew the final
5488 definition of the 'NULL type. This screwed up several key
5489 optimizations in the compiler, causing inefficiency in all sorts
5490 of places. (I found it because I wanted to understand why
5491 GET-INTERNAL-RUN-TIME was consing.)
5492 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5493 PROCLAIM FTYPE information. Unfortunately this broke binary
5494 compatibility again, since now the forms output by DEFGENERIC
5495 to refer to functions which didn't exist in 0.6.3.
5496 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5497 can use the new (as of 0.6.3) transform for SOME into MAP into
5499 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5500 values to (INTEGER 0 1000000), so that the time code will no longer
5501 occasionally get blown up by Linux returning 1000000 microseconds
5502 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5503 its output conform to the ANSI spec. (Alas, this makes its output
5504 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5505 * A full call to MAP NIL with a single sequence argument no longer conses.
5506 * fixes to problems pointed out by Martin Atzmueller:
5507 * The manual page no longer talks about multiprocessing as though
5508 it were currently supported.
5509 * The ILISP support patches have been removed from the distribution,
5510 because as of version 5.10.1, ILISP now supports SBCL without us
5511 having to maintain patches.
5512 * added a modified version of Raymond Toy's recent CMU CL patch for
5513 EQUALP comparison of HASH-TABLE
5515 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5517 * The system still can't cross-compile itself with
5518 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5519 entails), but at least it can compile more of itself that way
5520 than it used to be able to, and various buglets which were uncovered
5521 by trying to cross-compile itself that way have now been fixed.
5522 * This release breaks binary compatibility again. This time
5523 at least I've incremented the FASL file format version to 2, so that the
5524 problem can be detected reliably instead of just causing weird errors.
5525 * various new style warnings:
5526 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5527 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5528 ** using the lexical binding of a variable named in the *FOO* style
5529 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5530 as specified by ANSI.
5531 * *RANDOM-STATE* is no longer automatically initialized from
5532 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5533 default behavior of the system is to repeat its behavior every time
5534 it's run. If you'd like to change this behavior, you can always
5535 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5536 old convention there was no comparably easy way to get the system to
5537 repeat its behavior every time it was run.
5538 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5539 FTYPE declarations has been removed, in favor of their ANSI
5540 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5541 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5542 the types of their sequence arguments can be determined at compile time.
5543 This is done through a new open code expansion for MAP which eliminates
5544 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5545 argument types can be determined at compile time.
5546 * The optimizer now transforms COERCE into an identity operation when it
5547 can prove that the coerced object is already of the correct type. (This
5548 can be a win for machine generated code, including the output of other
5549 optimization transforms, such as the MAP transform above.)
5550 * Credit information has been moved from source file headers into CREDITS.
5551 * Source file headers have been made more standard.
5552 * The CASE macro now compiles without complaining even when it has
5555 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5557 * (Note that the way that the PCL macroexpansions were rewritten
5558 to accommodate the change in DEFGENERIC below breaks binary
5559 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5560 not run under sbcl-0.6.2. Once we get out of alpha releases,
5561 i.e. hit release 1.0.0, we'll probably try to maintain binary
5562 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5563 and sbcl-1.4.4. Until then, however, it might be fairly common
5564 for maintenance releases to break binary compatibility.)
5565 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5567 * The reporting of unhandled signals has been changed to print some
5568 explanatory text as well as the report form. (Previously only
5569 the report form was printed.)
5570 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5571 it defines, so that the compiler no longer issues undefined function
5572 warnings for compiled-but-not-yet-loaded generic functions.
5573 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5574 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5575 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5576 section of the ANSI spec are supported.
5577 * The "" nickname for the "KEYWORD" package has been removed.
5578 The reader still handles symbol tokens which begin with a package marker
5579 as keywords, but it doesn't expose its mechanism for doing so in the
5580 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5581 * The system now issues STYLE-WARNINGs for contradictory TYPE
5582 proclamations. (Warnings for contradictory FTYPE proclamations would
5583 be nice too, but those can't be done usefully unless the type system
5584 is made smarter about FUNCTION types.)
5585 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5586 systematized, so that "*target-*.lisp is supposed to exist only on the
5587 target and imply that there's a related file which exists on the
5588 host, and *host-*.lisp is supposed to exist only on the host and imply
5589 that there's a related file which exists on the target. This involves a
5590 lot of renaming. Hopefully the acute confusion caused by the renaming
5591 will be justified by the reduction in chronic confusion..
5592 ** runtime-type.lisp -> early-target-type.lisp
5593 ** target-type.lisp -> late-target-type.lisp
5594 ** early-host-format.lisp -> early-format.lisp
5595 ** late-host-format.lisp -> late-format.lisp
5596 ** host-error.lisp -> misc-error.lisp
5597 ** early-error.lisp -> early-target-error.lisp
5598 ** late-error.lisp -> late-target-error.lisp
5599 ** host-defboot.lisp -> early-defboot.lisp
5600 ** code/misc.lisp -> code/target-misc.lisp
5601 ** code/host-misc.lisp -> code/misc.lisp
5602 ** code/numbers.lisp -> code/target-numbers.lisp
5603 ** code/early-numbers.lisp -> numbers.lisp
5604 ** early-host-type.lisp -> early-type.lisp
5605 ** late-host-type.lisp -> late-type.lisp
5606 ** host-typep.lisp -> typep.lisp
5607 ** load.lisp -> target-load.lisp
5608 ** host-load.lisp -> load.lisp
5609 ** host-disassem.lisp -> disassem.lisp
5610 ** host-insts.lisp -> insts.lisp
5611 ** byte-comp.lisp -> target-byte-comp.lisp
5612 ** host-byte-comp.lisp -> byte-comp.lisp
5613 ** host-signal.lisp -> signal.lisp
5614 ** host-defstruct.lisp -> defstruct.lisp
5615 ** late-target-type.lisp -> deftypes-for-target.lisp
5616 Furthermore, several other previously target-only files foo.lisp (e.g.
5617 hash-table.lisp and random.lisp) have been split into a target-and-host
5618 foo.lisp file and a target-only target-foo.lisp file, with their key type
5619 definitions in the target-and-host part, so that the cross-compiler will
5620 know more about target types.
5621 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5622 gone away. In their place are various *BACKEND-FOO* variables
5623 corresponding to the slots of the old structure.
5624 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5625 into the target SBCL has been fixed.
5626 * The chill.lisp system for loading cold code into a running SBCL
5628 * Support for the CMU CL "scavenger hook" extension has been removed.
5629 (It was undocumented and unused in the CMU CL sources that SBCL was
5630 derived from, and stale in sbcl-0.6.1.)
5631 * Various errors in the cross-compiler type system were detected
5632 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5633 (enabling various consistency checks). Many of them were fixed,
5634 but some hard problems remain, so the compiler is back to
5635 running without *TYPE-SYSTEM-INITIALIZED* for now.
5636 * As part of the cross-compiler type system cleanup, I implemented
5637 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5638 * I have started adding UNCROSS calls throughout the type system
5639 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5640 on cross-compiler input files will be able to go away, and various
5642 * CONSTANTP now returns true for quoted forms (as explicitly required
5645 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5647 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5648 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5649 (MAKE-LIST -1). In the longer term, it ought to become true
5650 that declarations are assertions even at SAFETY 1. For now, it's not
5651 quite true even at SAFETY 3, but it's at least more nearly true..
5652 (Note that this change seems to increases the size of the system by
5653 O(5%) and to decrease the speed of the compiler by 20% or more.)
5654 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5655 for the problem of printing dozens of lines of distracting information
5656 about low-level system machinery as part of the top stack frame
5657 on entry to the debugger when an undefined function was called.
5658 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5659 is not reset to COMMON-LISP-USER.
5660 * Compilation of stuff related to dyncount.lisp has been made conditional
5661 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5662 smaller. The various dyncount-related symbols have been moved into
5663 a new "SB-DYNCOUNT" package.
5664 * tty-inspect.lisp has been renamed to inspect.lisp.
5665 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5666 feature has gone away. (When we eventually port to other flavors of
5667 libc and/or Unix, we'll try to make the differences between flavors
5668 invisible at the user level.)
5669 * Various other *FEATURES* tags, and/or their associated conditionals,
5670 have been removed if obsolescent, or given better documentation, or
5671 sometimes given more-mnemonic names.
5673 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5675 * tidied up "make.sh" script
5676 * tidied up system directory structure
5677 * better "clean.sh" behavior
5678 * added doc/FOR-CMUCL-DEVELOPERS
5679 * many many small tweaks to output format, e.g. removing possibly-confusing
5680 trailing #\. character in DESCRIBE-INSTANCE
5681 * (EQUALP #\A 'A) no longer signals an error.
5682 * new hashing code, including EQUALP hashing
5683 * tidied up Lisp initialization and toplevel
5684 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5685 * command line argument processing
5686 * added POSIX-GETENV function to deal with Unix-ish environment variables
5687 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5688 terminating SBCL on EOF
5689 * non-verbose GC by default
5690 * There is no more "sbcl" shell script; the sbcl file is now the C
5691 runtime executable (just like CMU CL).
5692 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5694 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5695 symbol and package information as Lisp native symbol and package objects
5696 instead of strings naming symbols and strings naming packages. This way,
5697 whenever packages are renamed (as in warm init), debug information is
5698 transformed along with everything else.
5699 * tweaked the optimization policy declarations which control the building
5700 of SBCL itself. Now, among other things, the system no longer saves
5701 source location debugging information. (This helps two problems at once
5702 by reducing SBCL size and by keeping SBCL from trying to look for its
5703 sources -- which may not exist -- when reporting errors.)
5704 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5705 debugging and testing purposes
5706 * cleaned up printing, making the printer call PRINT-OBJECT for
5707 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5708 methods, giving nearly-ANSI behavior
5709 * converted almost all special variables to use *FOO* naming convention
5710 * deleted PARSE-TIME functionality, since it can be done portably
5711 * moved some files out of cold init into warm init
5712 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5714 * regularized formatting of source files
5715 * added an install.sh script
5716 * fixed ridiculous memory usage of cross-compiler by making
5717 compiler/alloc.lisp not try to do pooling unless it can hook
5718 itself into the GC of the cross-compilation host. Now the system
5719 builds nicely on my old laptop.
5720 * added :SB-ALLOC in target-features.lisp-expr
5721 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5722 was not implemented)
5723 * re-did condition handling and note reporting in the compiler. Notes
5724 are no longer handled by signalling conditions. Style warnings
5725 and warnings are handled more correctly and reported in such a way
5726 that it's easy to find one or the other in your output (so that you
5727 can e.g. figure out which of many problems caused COMPILE-FILE to
5729 * changed the severity of several compiler warnings from full WARNING
5730 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5731 compiler note reporting so that it doesn't use the condition system
5732 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5733 COMPILE-FILE command)
5734 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5735 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5736 no longer supported, since their implementation was deeply intertwingled
5737 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5738 * removed bogus "support" for compiler macros named (SETF FOO), and
5739 removed the compiler macro for SETF INFO (but only after making a fool
5740 of myself on the cmucl-imp mailing list by posting a bogus patch for
5741 DEFINE-COMPILER-MACRO..)
5742 * Compiled files containing forms which have side effects on the Lisp
5743 reader (such as DEFPACKAGE forms) are now handled more correctly.
5744 (Compiler queuing of top level lambdas has been suppressed by setting
5745 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5746 * deleted various currently-unused source files, e.g. gengc.lisp. They
5747 may be added back at some point e.g. when porting to other architectures,
5748 but until they are it's distracting to distribute them and to try to
5750 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5751 there were so many of them they're just distractions, and UNCROSS is
5752 known to be able to handle the current sources
5753 * moved PROFILE functionality into TRACE, so that it will be clear
5754 how the wrapping and unwrapping of functions when you profile them
5755 interacts with the wrapping and unwrapping of functions when you
5756 trace them. (Actually, the functionality isn't there yet, but at least
5757 the interface specification is there. Hopefully, the functionality will
5758 arrive with some future maintenance release.)
5759 * removed host-oops.lisp
5760 * changed signature of QUIT function to allow UNIX-CODE argument
5761 * fixed READ-SEQUENCE bug
5762 * tweaked verbose GC output so that it looks more like the progress
5763 output that ANSI specifies for functions like LOAD
5764 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5765 * added <http://sbcl.sourceforge.com> to the banner information printed by