1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
3 * new contrib: sb-cover, an experimental code coverage tool, is included
5 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
7 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
8 produce more efficient code on x86-64 if the compiler can determine
9 that (+ INDEX <constant>) does not require a bounds check and FOO
10 has an element type at least 8 bits wide.
11 * enhancement: a new, experimental synchronous timeout facility is
12 provided. Refer to SB-SYS:WITH-DEADLINE for details.
13 * enhancement: when a symbol name conflict error arises, the
14 conflicting symbols are always printed with a package prefix.
15 (thanks to Kevin Reid)
16 * enhancement: stepping is now once again supported on the SPARC. (It is
17 also now more likely to work on CheneyGC builds on the PPC.)
18 * enhancement: Stepping support on MIPS.
19 * enhancement: sb-sprof can now also track and report accurate call
21 * incompatible change: PURIFY no longer copies the data from the
22 dynamic space into the static and read-only spaces on platforms
23 that use the generational garbage collector
24 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
26 * bug fix: GC race condition occasionally resulting in crashes with
27 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
29 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
31 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
32 a STYLE-WARNING to be signalled (regression from 1.0.4.)
33 * bug fix: an asynchronous interrupt could previously leave the
34 system running with GC inhibited.
35 * bug fix: a DECLARE form evaluated at top-level now causes an error
36 rather than silently (or verbosely) returning NIL.
37 * bug fix: trying to dissassemble functions compiled at high DEBUG could
38 result in a error being signalled due to source form lookup errors.
39 (reported by Peter Graves)
41 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
42 * incompatible change: removed writer methods for host-ent-name,
43 host-ent-addresses -- changing the values did not update the DNS
45 * minor incompatible change: changed experimental JOIN-THREAD interface
46 * documentation: the manual now lists reader and writer methods
47 in class slot documentation sections. (thanks to Richard M Kreuter)
48 * documentation: unwinding from asyncronous events has been
50 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
51 in multithreaded application code.
52 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
53 platforms. (thanks to James Anderson for the optimization hint)
54 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
56 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
57 variants no longer cons.
58 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
59 their NOT- variants no longer cons.
60 * optimization: EQUAL hash tables no longer use SXHASH for objects
61 of all data types, but instead use an EQL hash for types for which
62 EQUAL is the same as EQL.
63 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
64 are significantly faster.
65 * optimization: new STRING-OUTPUT-STREAM implementation is much
66 faster for large outputs and conses 30% less on average.
67 * enhancement: XREF information is now collected for references made
68 to global variables using SYMBOL-VALUE and a constant argument.
69 * enhancement: SIGINT now causes a specific condition
70 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
71 * bug fix: FILE-POSITION used to signal an error in some cases where
72 ANSI requires it to return NIL.
73 * bug fix: ADJUST-ARRAY is now interrupt-safe.
74 * bug fix: adding and removing fd-handlers is now interrupt-safe.
75 * bug fix: inlined calls to C now ensure 16byte stack alignment on
77 * bug fix: bad type declaration in the CLOS implementation has
78 been fixed. (reported by James Anderson)
79 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
80 (thanks to Lutz Euler)
81 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
82 required by the CLHS. (reported by Samium Gromoff)
83 * bug fix: dead unbound variable references now signal an error.
84 (reported by Marco Monteiro)
85 * bug fix: / with an unused value was being deleted in safe code.
86 (thanks to Marco Monteiro and Kevin Reid)
87 * bug fix: number of characters that can be written onto a single
88 line in a file is unlimited.
89 * bug fix: some GC deadlocks caused by asynchronous interrupts have
90 been fixed by inhibiting interrupts for when GC is disbled.
91 * bug fix: some interrupt safety issues with GETHASH, (SETF
92 GETHASH), CLRHASH and REMHASH have been fixed.
93 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
94 specifier no longer causes infinite recursion.
95 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
96 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
97 * bug fix: an &environment argument with the correct variable information
98 is passed to macros that are expanded during byte compilation.
99 (reported by Samium Gromoff)
100 * bug fix: a logic error was causing the hash function for bignums to
101 have excessive amounts of collisions. (reported by Faré Rideau)
102 * bug fix: modifying the contents of an array could change the return
103 value of SXHASH on that array, which is only allowed for strings
104 and bit vectors. (bug introduced in 0.9.16)
105 * improvement: the x86-64/darwin port now passes all tests (except
106 for the debugger tests) but should still be considered
107 experimental until this is fixed.
108 * improvement: a style-warning is signaled for CASE (etc) clauses with
109 duplicate keys. (thanks to Kevin Reid)
110 * improvement: macos/x86 and macos/x86-64 now use mach exception
111 handlers for dealing with illegal instructions (for trapping and
112 error handling) and memory protection violations (for GC).
114 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
115 * new platform: experimental support for x86-64/darwin (MacOS).
116 * incompatible change: the thread-safe (on most platforms) getaddrinfo
117 and getnameinfo sockets functions are used instead of gethostbyaddr
118 and gethostbyname, on platforms where the newer functions are available.
119 As a result, the ALIASES field of HOST-ENT will always be NIL on these
121 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
122 don't print the startup banner, but behave as if --noinform was passed
123 as a command line argument. (thanks to Kevin Reid)
124 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
125 * optimization: code using alien values with undeclared types is much faster.
126 * optimization: the compiler is now able to open code SEARCH in more cases.
127 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
128 * bug fix: using standardized COMMON-LISP special variables as loop
129 variables no longer signals bogus package lock violations. (reported
131 * bug fix: declaring local loop variables to be of a range-limited type
132 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
133 (reported by Andras Simon)
134 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
135 bugs remain on x86-64.)
136 * bug fix: the #= and ## reader macros now interact reasonably with
137 funcallable instances.
138 * bug fix: type-checks for function arguments were compiled using the
139 compiler policy settings of the wrong lexical environment (in 1.0.2
141 * bug fix: SHADOW accepts characters as string designators, as required
142 by the spec (thanks to Eric Marsden)
143 * bug fix: fixed GC safety issues when foreign functions are called with
144 non-base strings as arguments
145 * bug fix: more consistent error output for fatal-compiler-warnings, like
147 * bug fix: functions with proper names were showing up as NIL in
148 backtraces (reported by Edi Weitz, regression in 1.0.2)
150 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
151 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
152 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
153 produce more efficient code on the x86 if the compiler can determine
154 that (+ INDEX <constant>) does not require a bounds check and FOO
155 has an element type at least 8 bits wide.
156 * bug fix: references to undefined variables are handled the same way
157 in toplevel forms as in the normal compiler
158 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
160 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
161 (thanks to Jon Buller)
162 * improvement: faster compilation times for complex functions
163 * improvement: added readlink support to SB-POSIX (thanks to Richard
166 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
167 * improvement: experimental support for mach exception handling on
168 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
170 * improvement: support for GBK external format.
171 (thanks to Chun Tian (binghe))
172 * improvement: the debugger now displays variables that have been closed
173 over, in code compiled with (DEBUG 2) or higher
174 * improvement: support for executable cores on NetBSD (thanks to
176 * new feature: added a RESTART-FRAME debugger command
177 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
178 be used to provide an implementation for FILE-POSITION on Gray streams
179 (thanks to Eric Marsden)
180 * new feature: add syslog, openlog and closelog support to SB-POSIX
181 (thanks to Richard Kreuter)
182 * optimization: the function call overhead in code compiled with
183 a high DEBUG optimization setting is significantly
184 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
185 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
186 (thanks to Tony Martinez)
187 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
188 evaluated once (reported by Kevin Reid)
189 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
190 works again (thanks to Kilian Sprotte)
191 * bug fix: an error is signaled for tagbodies with duplicate tags
192 (thanks to Stephen Wilson)
193 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
194 * bug fix: Win32 port can now handle foreign code unwinding Lisp
195 stack frames from alien callbacks.
196 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
197 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
198 65k pages (thanks to David Woodhouse)
199 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
201 changes in sbcl-1.0.1 relative to sbcl-1.0:
202 * new platform: FreeBSD/x86-64, including support for threading.
203 * new feature: the compiler stores cross-referencing information
204 abount function calls (who-calls), macroexpansion (who-macroexpands)
205 and special variables (who-binds, who-sets, who-references) for code
206 compiled with (< SPACE 3). This information is available through the
207 sb-introspect contrib.
208 * new feature: users may subclass SEQUENCE, and have instances of
209 these classes interoperate with standard Common Lisp functions if
210 a number of methods are defined. (This feature is experimental
211 and the interface subject to change based on feedback from SBCL
212 users and the general community)
213 * improvement: sb-sprof traces call stacks to an arbitrary depth on
214 x86/x86-64, rather than the previous fixed depth of 8
215 * improvement: another pthread back-end of mutex "pthread-futex".
216 * bug fix: non-ascii command-line arguments are processed correctly
217 (thanks to Yaroslav Kavenchuk)
218 * bug fix: non-required arguments were not passed correctly when a method
219 defined using DEFMETHOD was called from a mop-generated method using
220 CALL-NEXT-METHOD (reported by Pascal Costanza)
221 * bug fix: recursion is now permitted in accessors through
222 SLOT-UNBOUND. (reported by Pascal Costanza)
223 * bug fix: an error was signaled at startup if the HOME environment
224 variable was defined, but had an empty value (reported by Peter Van Eynde)
225 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
226 signal a PROGRAM-ERROR, not a BUG.
227 * bug fix: SB-POSIX:READDIR works when built with large file support.
228 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
230 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
231 for better automatic indentation support. (thanks to Matt Pillsbury)
232 * optimization: loading generic functions no longer takes O(n^2) time,
233 proportional to the amount of methods in the generic function
234 (reported by Todd Sabin and Jeremy Brown)
235 * optimization: the FIND and POSITION family of sequence functions
236 are significantly faster on arrays whose element types have been
238 * improvements to the Windows port:
239 ** Intermittent heap corruption problems have been fixed. (thanks
240 to Alastair Bridgewater)
241 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
243 ** Lisp is able to unwind foreign exception frames from alien
244 callbacks. (thanks to Alastair Bridgewater)
246 changes in sbcl-1.0 relative to sbcl-0.9.18:
247 * improvement: experimental support for threading on FreeBSD/x86.
248 (thanks to NIIMI Satoshi)
249 * improvement: runtime option --dynamic-space-size can be used
250 to set the size of the dynamic space reserved on startup.
251 * improvement: floating point modes in effect are now saved in
252 core, and restored on startup.
253 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
254 startup, not time since first call to GET-INTERNAL-REAL-TIME.
255 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
256 threads are not running after *SAVE-HOOKS* have run.
257 * improvement: writes to CLOS instance slots are type-checked in code
258 compiled with (SAFETY 3)
259 * improvement: floating-point exception handling on FreeBSD (thanks to
261 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
262 (thanks to Zach Beane)
263 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
265 * improvement: added support for the Shift-JIS external format.
266 (contributed by NIIMI Satoshi)
267 * improvement: callbacks are supported on Linux/PPC. (thanks to
269 * bug fix: compiler bug triggered by a (non-standard) VALUES
270 declaration in a LET* was fixed. (reported by Kaersten Poeck)
271 * bug fix: file compiler no longer confuses validated and already
272 dumped structurres. (reported by Kaersten Poeck)
273 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
274 fill-pointer signals a type-error as required. (thanks to
276 * bug fix: disassemly of funcallable instances works.
277 * bug fix: single stepping on PPC.
278 * bug fix: fix thread-safety problems in the type system (generally
279 manifesting as nonsensical errors like "STRING is a bad type specifier
280 for sequences" or "The value 1 is not of type FIXNUM")
281 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
282 (reported by Josip Gracin)
283 * bug fix: an error is signaled for attempts to displace arrays with
284 incompatible element types (thanks to Mario Mommer)
285 * bug fix: more correct handling of wide characters in debug info
286 (bug reported by Attila Lendvai and fixed by Juho Snellman)
287 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
288 and don't cause extra consing
289 * optimization: MAP and MAP-INTO are significantly faster on vectors
290 whose elements types have been declared.
291 * Improvements to SB-SPROF:
292 ** Support for allocation profiling
293 ** Reduced profiling overhead, especially for long profiling runs
294 * Improvements to the Windows port:
295 ** floating point exceptions are now reported correctly.
296 ** stack exhaustion detection works partially.
297 ** more accurate GET-INTERNAL-REAL-TIME.
298 ** better RUN-PROGRAM behaviour in case of errors in the child
300 ** PROBE-FILE now simplifies pathnames correctly.
301 ** DIRECTORY now works correctly with :WILD-INFERIORS.
302 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
303 1st 1970 (although time timezone data will be unreliable).
304 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
306 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
307 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
308 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
310 * bug fix: two potential GC deadlocks affecting threaded builds.
311 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
312 returns T (reported by Anton Kazennikov)
313 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
314 with non-variable places
315 * bug fix: remove a race condition in the setting of
316 funcallable-instance functions, this should make threaded CLOS
317 code more stable against memory faults.
318 * bug fix: corruption of specials when unbinding is interrupted by an
319 asynchronous unwind (reported by Hannu Koivisto)
320 * improvement: the debugger will now also display local variables that
321 are only used once, for code compiled with a DEBUG optimization quality
324 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
325 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
326 * incompatible change: External-format support for FFI calls. The
327 SB-ALIEN:C-STRING no longer implies an ASCII
328 external-format. Instead the string is subject to external-format
329 conversion. Additionally return values of type SB-ALIEN:C-STRING
330 are by default of type (SIMPLE-ARRAY CHARACTER), not
331 SIMPLE-BASE-STRING. For an alien type that behaves like the old
332 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
333 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
334 * incompatible change: SB-EXT package no longer contains the
335 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
336 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
338 * incompatible change: the single-stepper is no longer available
339 on Alpha, Mips and Sparc platforms.
340 * minor incompatible change: the direct superclasses of
341 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
342 not (STANDARD-OBJECT FUNCTION). This makes the
343 class-precedence-lists of GENERIC-FUNCTION and
344 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
346 * minor incompatible change: the default stream external format on
347 non-unicode SBCL is no longer detected from the locale, but is always
349 * new feature: Add a version of evaluator that uses an interpreter instead
350 of the compiler. EVAL still uses the compiler by default, to switch it
351 to use the interpreter, set the value of the variable
352 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
353 * minor incompatible change: the single-stepper REPL has been merged
354 with the normal debugger (see the "Stepping" heading of the debugger help
355 for more details). The debugger command STEP will no longer switch
356 to the single-stepper REPL.
357 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
358 for a type now works.
359 * bug fix: loading of default sysinit file works. (thanks to Leonid
361 * bug fix: better detection of circularities in the file-compiler.
362 (reported by Marco Monteiro)
363 * bug fix: the CL pathname functions now work with files that have
364 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
365 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
366 systems with Unix98 pty semantics.
367 * bug fix: ASDF-INSTALL will now work with bsd tar.
368 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
370 * bug fix: timers expiring in dead threads no longer cause a
371 type-error (reported by Paul "Nonny Mouse").
372 * bug fix: thanks to more lightweight single-stepper instrumentation,
373 code compiled with (DEBUG 3) will compile and execute significantly faster,
374 and will have more accurate type-inferencing than before
375 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
376 whose bindings are modified
377 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
378 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
379 ** replace dummy implementations of CL:MACHINE-INSTANCE and
380 CL:SOFTWARE-VERSION with proper version
382 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
383 * feature: implemented the READER-METHOD-CLASS and
384 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
385 as specified by AMOP.
386 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
388 * optimization: faster LOGCOUNT implementation on x86 and x86-64
389 (thanks to Lutz Euler)
390 * optimization: hashing of general arrays and vectors has been
391 improved. (reported by Any Fingerhut)
392 * enhancement: SB-INTROSPECT is now able to find definitions of
393 profiled functions. (thanks to Troels Henriksen)
394 * enhancement: compiler-macro expansion applies now to FUNCALL forms
395 as well, allowing compiler-macros for SETF-functions to expand.
396 * enhancement: step-instrumentation no longer wraps known
397 single-value functions with multiple-value context, allowing
398 better type inference.
399 * fixed bug #337: use of MAKE-METHOD in method combination now works
400 even in the presence of user-defined method classes. (reported by
401 Bruno Haible and Pascal Costanza)
402 * fixed bug #339(c): if there are applicable methods not part of any
403 long-form method-combination group, call INVALID-METHOD-ERROR.
404 (reported by Bruno Haible)
405 * fixed bug #361: the :FUNCTION initarg in the protocol for
406 initialization of methods can now be used to override
407 internally-produced optimized functions. (reported by Bruno
409 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
410 system-provided lambda expression no longer cause warnings about
411 unbound #:|pv-table| symbols.
412 * bug fix: improved the handling of type declarations and the
413 detection of violations for keyword arguments with non-constant
415 * bug fix: potentially erronous calls to PATHNAME and
416 MERGE-PATHNAMES were being flushed in some cases.
417 (reported by Richard Kreuter)
418 * bug fix: compiled calls to TYPEP were mishandling obsolete
419 instances. (reported by James Bielman and Attila Lendvai)
420 * bug fix: format strings with ~> without matching ~< no longer
421 trigger an AVER, but signal an understandable error instead.
422 (reported by Antonio Martinez)
423 * bug fix: specifying an output-file without a pathname-type for
424 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
425 of input-file instead of "fasl". (reported by Robert Dodier)
426 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
427 (reported by James Y Knight).
428 * bug fix: compiler-macros-function did not consider the environment
429 argument for shadowing by local functions.
430 * bug fix: compiler-macros expansion was inhibited by local INLINE
432 * bug fix: inline expansions of known functions were subject to
433 step-instrumentation in high debug policies, leading to problems
435 * bug fix: compiler failed to differentiate between different CONS
437 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
438 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
439 * bug fix: #368: incorrect use of expressed vs. upgraded array
441 * bug fix: #306a: more precise unions of array types.
442 * thread-safety improvements:
443 ** CONDITION-WAIT could return early on Linux, if the thread was
444 interrupted and subsequently continued with SIGCONT.
445 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
447 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
448 * added support for the ucs-2 external format. (contributed by Ivan
450 * minor incompatible change: pretty printing of objects of type
451 (cons symbol) is, in the default pprint-dispatch-table, now
452 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
454 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
455 called later in a class's lifetime, possibly as late as when the
456 first instance of the class is created. Previously,
457 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
458 class became finalizeable.
459 * fixed bug: FILE-POSITION sometimes returned inconsistent results
460 for multibyte external-format streams. (thanks to "vbzoli")
461 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
462 with :ALLOCATION :CLASS inherited from superclasses of the
464 * fixed bug: anonymous classes can now be created using the :NAME
465 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
466 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
467 * fixed bug: core-files saved with :EXECUTABLE T can again be
468 executed when SBCL_HOME isn't set. (reported by James Knight)
469 * fixed bug: toplevel LOCALLY forms with declarations could
470 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
471 * fixed bug: printing from several different threads using different
472 values of *print-case* could cause invalid output, due to
473 some internal special variables of the printer not being bound
474 thread-locally (reported by Max Mikhanosha)
475 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
476 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
477 (reported by Pascal Costanza)
478 * fixed bug: FINALIZE-INHERITANCE is called from
479 REINITIALIZE-INSTANCE on classes when the class has previously
480 been finalized, as required by AMOP.
481 * minor code generation optimizations:
482 ** better register allocation in CLOS dispatching functions
483 ** overflow detection when coercing signed bytes to fixnums on x86-64
484 ** is now implemented with one IMUL instruction instead of three shifts
485 ** more efficient bit-vector access on x86 and x86-64
486 ** more efficient access to raw structure slots on x86 and x86-64
487 * fixed some bugs revealed by Paul Dietz' test suite:
488 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
491 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
492 * feature: thread support on Solaris/x86, and experimental thread support
494 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
496 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
497 default method for SLOT-UNBOUND.
498 * minor incompatible change: prevent the user from specializing the
499 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
500 somewhat counter to the intent of the protocol, I (CSR) think, and
501 additionally it just doesn't work in SBCL as currently
502 implemented, thanks to optimizations (that are always valid for
503 the other three SLOT-VALUEish functions, but not for the setter).
504 * minor incompatible change: the :SB-LDB feature is now enabled by
505 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
506 the low-level debugger.
507 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
508 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
509 down to an SB-SYS:FD-STREAM, allowing direct communication in
510 more cases. Temporary files and pipes are still used as a fallback
512 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
513 on the non-reentrant C functions localtime(3) and gmtime(3).
514 * bug fix: global optimization policy was not visible in LOCALLY and
516 * bug fix: class objects can be used as specializers in methods.
517 (reported by Pascal Costanza)
518 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
519 with a type error. (reported by blitz_ on #lisp)
520 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
521 failed with a type error. (reported by Pascal Bourguignon)
522 * bug fix: merging pathnames against defaults with :DIRECTORY
523 starting with '(:RELATIVE :BACK) should preserve the :BACK.
524 (reported by James Y Knight)
525 * bug fix: saving large (>2GB) cores on x86-64 now works
526 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
527 constant index and a (simple-array (signed-byte 32)) array
528 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
529 error (patch by Robert J. Macomber)
530 * feature: TIME now displays the amount of run-time spent in GC
531 * bug fix: The debugger now does a better job of respecting
532 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
533 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
534 workaround for bug 403.)
535 * bug fix: floating point exception handling now works on FreeBSD
536 (thanks to NIIMI Satoshi)
537 * fixed some bugs revealed by Paul Dietz' test suite:
538 ** MISC.641: LET-conversion were not supposed to work in late
540 * improvements to the Win32/x86 port:
541 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
542 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
544 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
545 * new feature: source path information is generated for macro-expansion
546 errors for use in IDE's like Slime (thanks to Helmut Eller)
547 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
548 * bug fix: compiler does not loop forever on an invalid type in
550 * improvement: compilation of most CLOS applications is significantly
552 * optimization: added a limited bytecode compiler for simple toplevel
553 forms, speeding up compilation and FASL loading
554 * bug fix: the statistical profiler now properly distinguishes anonymous
557 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
558 * minor incompatible change: in sbcl-0.9.11 (but not earlier
559 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
560 into subclasses of the finalized class. Now user calls to
561 FINALIZE-INHERITANCE finalize just the one class, and calls by the
562 system return before any subclasses are finalized.
563 * minor incompatible change: The reader no longer ignores errors
564 regarding non-existent packages in #+ and #- feature tests.
565 * new feature: command line options --no-sysinit, --no-userinit to
566 inhibit loading the corresponding init files
567 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
568 for sending data through UDP sockets (thanks to François-René Rideau)
569 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
570 error is signalled instead (thanks to François-René Rideau)
571 * minor incompatible change: Error signalling behaviour of lexical
572 operations violating package locks has changed slightly. Refer to
573 documentation on package locks for details.
574 * enhancement: EVAL can process IF-expressions without resorting to the
576 * bug fix: Some operations involving SIN, COS, or TAN weren't being
577 constant-folded properly due to missing out-of-line functions.
578 (reported by Mika Pihlajamäki)
579 * bug fix: LISTEN sometimes returned T even in cases where no data was
580 immediately available from the stream
581 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
582 were reversed. (reported by Levente Mészáros)
583 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
584 when given a vector displaced to an adjustable array. (reported
586 * bug fix: derivation of float boundaries from numbers outside the
587 appropriate float range (reported by John Wiseman)
588 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
590 * bug fix: precomputing cache entries for generic functions with
591 some subclasses of specializers as yet invalid does not attempt to
592 fill a cache line with a negative offset. (reported by Levente
594 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
595 allow condition class objects as arguments to DOCUMENTATION and
596 (SETF DOCUMENTATION); only find and set documentation for
597 structure names for the STRUCTURE doc-type. (suggested by Gary
599 * improvements to the Win32/x86 port:
600 ** (user-homedir-pathname) and default initialization file
601 locations now know about the user's "Documents and Settings"
602 directory (thanks to Yaroslav Kavenchuk)
603 ** run-program is implemented (thanks to Mike Thomas)
604 ** sockets support (thanks to Timothy Ritchey)
605 ** better backtrace support (thanks to Alastair Bridgewater)
606 ** sb-grovel supported
607 ** asdf-install and sb-posix work somewhat
608 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
609 * improvements to the Solaris/x86 port:
610 ** works on Solaris 11/Solaris Express
611 ** floating-point exception handling support
612 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
613 * fixed some bugs revealed by Paul Dietz' test suite:
614 ** REMOVE-METHOD returns its generic function argument even when
615 the method is not one of the generic functions' methods.
616 ** objects declared as MEMBER types can be admissible arguments to
618 ** printing characters should simply be printed by the FORMAT ~:C
620 ** compiler failure when compiling functions with hairy constant
621 defaults for optional parameters.
622 ** compiler produces wrong code when MAYBE-INLINE-expanding a
623 function, which is already optimized.
625 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
626 * new platform: experimental support for SBCL x86/Darwin, including
627 MacOS X 10.4.5 on Intel.
628 * new feature: Unicode character names are now known to the system
629 (through CHAR-NAME and NAME-CHAR).
630 * new feature: the filesystem location of SBCL's core file is
631 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
632 * minor incompatible change: the contrib modules SB-POSIX and
633 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
634 this change is to make it easier to distribute
635 location-independent binaries.
636 * bug fix: as implied by AMOP, standardized classes no longer have
637 slots named by external symbols of public packages. (reported by
639 * optimization: calling functions via constant symbols -- as in
640 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
641 via the function object as in (FUNCALL #'FOO).
642 * optimization: on x86 and x86-64, the calling convention for the
643 case of dealing with an unknown number of values has been altered
644 to be friendlier to the prediction heuristics implemented,
645 particularly with reference to CALL and RET pairing. (thanks to
646 Alastair Bridgewater)
647 * optimization: on x86, the code for access to thread-local symbol
648 values and binding/unbinding of thread-local symbols is smaller.
649 (thanks to Alastair Bridgewater)
650 * enhancement: CONSTANTP is now able to determine constantness of
651 more complex forms, including calls to constant-foldable standardized
652 functions and some special forms beyond QUOTE.
653 * fixed bug: occasional GC crashes on Solaris/x86
654 * optimization: x86-64 supports stack allocation of results of simple
655 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
656 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
657 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
658 (thanks to James Knight)
659 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
660 by the standard (thanks to Stephen Compall)
662 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
663 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
664 be used for bundling the runtime and the core file into one
665 executable binary. This feature is not currently supported on all SBCL
666 platforms. (thanks to James Bielman and NIIMI Satoshi)
667 * new feature: a generational or ephemeral garbage collector is now
668 the default on the PowerPC platform (both Linux and Darwin). The
669 old Cheney (stop and copy) collector is a build-time option.
670 (thanks to Cyrus Harmon, after Raymond Toy)
671 * minor incompatible change: the method by which SBCL finds its
672 contributed modules has changed; it no longer relies on symbolic
673 links from an $SBCL_HOME/systems directory, but searches directly
674 in the subdirectories of $SBCL_HOME.
675 * enhancement: the dynamic heap size on the Linux/PPC platform is
676 markedly larger, even using the older Cheney garbage collector.
677 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
678 certain complicated string types. (reported by Gary King)
679 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
680 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
681 * fixed bug: DOCUMENTATION on structures no longer signals an
682 error if no structure type of the right name exists. (reported by
684 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
685 applicable non-standard primary or :AROUND method on
686 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
687 no longer cause unbound variable errors. (reported by Kevin Reid)
688 * fixed bug: metacircle resolution in cases where methods have slots
689 added before the slots from STANDARD-METHOD. (reported by Jean
691 * fixed bug: the Power PC assembly code for calling into lisp
692 sometimes computed its return address wrongly (depending on the
693 15th bit of the address, fixed at link time). This used to
694 manifest itself in a segmentation violation while building PCL.
695 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
696 many others over the years)
697 * fixed bug: ROOM no longer reports silly numbers for stack usage if
698 the stack pointer (treated as a signed integer) is negative.
699 (thanks to Peter van Eynde)
701 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
702 * new platform: experimental support for the Windows operating
703 system has been added. (thanks to Alastair Bridgewater)
704 * fixed several bugs in and robustified the PPC FFI (including
705 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
706 * bug fix: allow non-simple string symbol names (reported by Paul
708 * bug fix: interrupt handling on NetBSD (thanks to Richard M
710 * bug fix: saving a core corrupted callbacks on x86/x86-64
711 * bug fix: closed a loophole in metacircularity detection and
712 grounding in the PCL implementation of CLOS.
713 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
714 (reported by tomppa on #lisp)
715 * bug fix: clear the direction flag on Lisp -> C transitions, as
716 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
717 (reported by Andrej Grozin and Hendrik Maryns)
718 * optimization: major improvements to GC efficiency on GENCGC platforms
719 * optimization: faster implementation of EQUAL
720 * optimization: emit more efficient opcodes for some common
721 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
722 * optimization: several other minor code-generation improvements on x86-64
723 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
725 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
726 * minor incompatible change: (SETF CLASS-NAME) and (SETF
727 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
728 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
729 by ANSI as a generic function, and for consistency (SETF
730 GENERIC-FUNCTION-NAME) is treated likewise.
731 * fixed bug #233.b: make constraint propagation notice when a variable
732 value is changed after it is referenced but before it is used
733 * fixed bug #296: no more arbitrary behaviour from filenames with
734 odd characters as --load arguments. (reported by Adam Warner)
735 * fixed bug #390: :CHARACTER-SET pathname components now work as
736 expected. (reported by Tim Daly Jr)
737 * fixed bug #391: complicated :TYPE intersections in slot
738 definitions no longer cause an error in PCL internals.
739 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
740 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
741 returning the number of octets which would be written to the
742 file-stream. (thanks to Robert J. Macomber)
743 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
744 arguments to the internal stream functions. (thanks to David
746 * bug fix: stack exhaustion is now handled on the x86/SunOS
747 platform. (thanks to Daisuke Homma)
748 * bug fix: more accurate ROOM results on GENCGC platforms
749 * optimization: improved type inference for arithmetic-for
750 index variables in LOOP
751 * optimization: faster floating-point SQRT on x86-64
752 * fixed some bugs revealed by Paul Dietz' test suite:
753 ** DOCUMENTATION returns NIL instead of "" for method combinations
754 that don't have a docstring
756 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
757 * minor incompatible change: (SETF CLASS-NAME) and (SETF
758 GENERIC-FUNCTION-NAME) are no longer generic functions, and
759 therefore are not customizeable by user code (as seems to be at
760 least permitted and maybe required by AMOP). As a consolation,
761 however, the SBCL implementation of these functions now calls
762 REINITIALIZE-INSTANCE as specified by AMOP.
763 * bug fix: slot-definition documentation is propagated as per ANSI
764 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
765 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
766 argument, as do eslotd initialization methods. (from Pascal
767 Costanza's "Closer" project)
768 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
769 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
771 * bug fix: it is now possible to have more than one subclass of
772 STANDARD-GENERIC-FUNCTION without causing stack overflow.
773 (reported by Bruno Haible, Pascal Costanza and others)
774 * bug fix: the dependent update protocol now works for generic
775 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
777 * bug fix: condition-class instances corresponding to
778 DEFINE-CONDITION forms are now created eagerly. (reported by
779 Kalle Olavi Niemitalo on comp.lang.lisp)
780 * bug fix: floating point printing is more accurate in some
781 circumstances. (thanks to Simon Alexander)
782 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
783 merged with *DEFAULT-PATHNAME-DEFAULTS*.
784 * bug fix: callbacks on OS X now preserve stack-alignment.
785 (thanks to Cyrus Harmon)
786 * enhancement: the x86-64 disassembler is much better at
787 disassembling SSE instructions. (thanks to Lutz Euler)
788 * enhancement: CHECK-TYPE on a variable now allows type-inference
789 (thanks to Vincent Arkesteijn)
790 * optimization: improved performance of EUC-JP external format.
791 (thanks to NIIMI Satoshi)
792 * optimization: performance improvements to IO on file streams of
793 :ELEMENT-TYPE CHARACTER
794 * optimization: much faster memory allocation on x86-64
795 * optimization: faster garbage collections (latency and throughput)
797 * optimization: faster bignum subtraction and division on x86 and x86-64
798 (thanks to Lutz Euler)
799 * optimization: more accurate type-inference for loops that use a
800 floating point index variable or a negative step.
802 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
803 * bug fix: add a workaround to SBCL looping infinitely at startup on
804 Linux kernels with apparently buggy implementations of personality().
805 (thanks to Svein Ove Aas)
806 * bug fix: Unicode symbols are correctly printed in LDB backtraces
807 (thanks to David Lichteblau)
808 * bug fix: local bindings shadow global symbol macros for
809 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
810 * optimization: non-open coded uses of numeric comparison operators
811 (e.g. >) no longer cons when called with more than one parameter
812 on platforms supporting dynamic-extent allocation.
813 * enhancement: saving cores with foreign code loaded is now supported
814 on MIPS/Linux in addition to the previously supported platforms.
815 * bug fix: FIND-RESTART now tests for activity, not applicability when given
816 a restart object as identifier. (reported by Helmut Eller for CMUCL)
817 * bug fix: division by zero in sb-sprof when no samples were collected
818 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
819 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
820 (thanks to Alexander Kjeldaas)
821 * new feature: ignoring the return values of destructive functions
822 like NREVERSE signals a compile-time style-warning.
823 (thanks to Kevin Reid)
824 * new feature: the alignment of alien structure fields can be
825 explicitly specified. (thanks to Cyrus Harmon)
826 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
827 and dump core on SIGQUIT
829 ** incompatible change: threads do not inherit values of specials
830 from their parents (see manual)
831 ** bug fix: threads stacks belonging to dead threads are freed by the
832 next exiting thread, no need to gc to collect thread stacks anymore
833 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
834 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
835 ** bug fix: dynamic variable and thread start related gc lossage
836 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
838 * fixed some bugs revealed by Paul Dietz' test suite:
839 ** SUBTYPEP is slightly more accurate on heinously complicated
840 CONS types where some of the members have uncertain (in the
841 NIL, NIL sense) type relationships to each other.
843 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
844 no longer purified unless :PURIFY T is explicitly specified.
845 ** Non-purified cores are significantly smaller than before
847 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
848 * new feature: timers based on Zach Beane's excellent timer package
849 * added support for the following external formats: koi8-u,
850 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
851 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
852 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
853 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
854 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
855 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
856 * incompatible change: a threaded SBCL will no longer revert to
857 non-threaded mode on non-NPTL systems, but refuse to start entirely.
858 * bug fix: interrupts are disabled until startup is complete; no
859 more sigsegvs when receiving a signal to soon
860 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
862 * bug fix: add a workaround for the memory randomization features in
863 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
864 workaround will only be in effect on systems with the proc filesystem
866 * bug fix: printing objects of type HASH-TABLE signals a
867 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
869 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
870 have been running for over 50 days. (reported by Gilbert Baumann)
871 * bug fix: the logic for getting names of functions gets less
872 confused when confronded with alternate-metaclass
873 funcallable-instances. (reported by Cyrus Harmon)
874 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
875 consistent, even on internal alternate-metaclass objects.
876 * bug fix: flush closure information collected by physical
877 environment analysis prepass before the main pass. (bug reported
879 * bug fix: compiler pack phase does not modify a hashtable, which is
880 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
881 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
882 may be expanded. (reported by Alexander <malishev> on c.l.l)
883 * bug fix: no more style-warnings when compiling code using LAST
884 under high-SPEED low-SPACE settings. (reported by David Wallin
886 * fixed some incompatibilities between SBCL's MOP and the MOP
888 ** the METAOBJECT class is now implemented;
889 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
890 STANDARD-OBJECT, as required;
891 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
892 now compatible (as required by VALIDATE-SUPERCLASS); there
893 remains a consistency requirement over the presence or absence
894 of the FUNCTION class in the superclass list of finalized
895 classes; see the manual for more details;
896 ** the :AROUND method for COMPUTE-SLOTS on
897 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
898 requested slot ordering.
900 ** bug fix: parent thread now can be gc'ed even with a live
902 ** bug fix: binding a special with PROGV to no value is not
903 the same as that symbol not having been bound (thanks to
905 ** bug fix: binding specials is thread safe (thanks to
907 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
908 and SB-SPROF do not die with 'no handler for signal XX in
909 interrupt_handle_now(..)' anymore
910 ** bug fix: WITH-TIMEOUT works with multiple running threads
911 * fixed some bugs revealed by Paul Dietz' test suite:
912 ** ENSURE-GENERIC-FUNCTION should take a method class object for
913 the :method-class keyword argument.
915 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
916 * new port: the Solaris operating system on x86 processors is now
917 mostly supported, though some rough edges in the environment
918 remain. (thanks to Daisuke Homma)
919 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
920 space for its heap. (thanks to Thiemo Seufer)
921 * optimization: SBCL on MIPS platforms now supports dynamic-extent
922 closures. (thanks to Thiemo Seufer)
923 * minor incompatible change: eof selects abort in the debugger.
924 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
925 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
927 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
928 deprecated, and will go away in a future revision of SBCL.
929 * minor incompatible change: GC-ON and GC-OFF are no longer
930 implemented with a counter, it does not matter how many times gc
931 is switched on or off
932 * bug fix: discriminating functions for generic function classes
933 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
934 make invalid assumptions about method precedence order. (reported
936 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
937 platforms. (thanks to Thiemo Seufer)
938 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
939 * bug fix: correct pathname printing with printer escaping is on.
940 (thanks to Kevin Reid)
941 * bug fix: complex VOP definitions in "user-space" no longer trigger
942 package locks. (reported by Zach Beane)
943 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
944 for funcallable-instance functions, and is no different from
946 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
947 characters in the current readtable. (reported by Nicholas Neuss)
948 * bug fix: SUBTYPEP on various CONS types returns more a more
949 accurate acknowledgment of its certainty.
950 * optimizations: REMOVE-DUPLICATES now runs in linear time on
951 lists in some cases. This partially fixes bug 384.
952 * flush all standard streams before prompting in the REPL and the
954 * bug fix: signal handling and triggering gc do not conflict
955 directly anymore, in particular a high frequency sb-sprof does
956 not prevent gc from running
957 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
958 approximation for timezone and DST information between the
959 universal time epoch and the smallest negative 32-bit time_t.
960 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
961 year 1899 so long as the encoded time is non-negative.
962 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
963 the correct answer for non-constant rotate amounts when performing
964 an inline 32-bit rotation.
966 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
967 there is only one thread in the session
968 ** bug fix: memory leak for streams created in one thread and
969 written to in another
970 ** bug fix: lockup when compiled with gcc4
971 ** bug fix: race that allows the gc to be triggered when gc is
973 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
974 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
976 ** bug fix: finished threads are gc'ed properly
977 * fixed some bugs revealed by Paul Dietz' test suite:
978 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
979 has arguments with a different set of applicable methods from
980 the orignal arguments.
981 ** The type error thrown by MAP now has a correct expected-type
983 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
984 name a compiled function.
985 ** Three MISC tests where a large bignum was improperly coerced to
986 a float that couldn't represent that bignum during type
987 derivation were fixed.
988 ** SUBTYPEP can now handle types involving the negation of a
989 list-form FUNCTION type.
990 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
991 as (COMPLEX (AND RATIO (NOT FIXNUM))).
992 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
994 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
995 * New feature: Experimental support for bivalent streams: streams
996 opened with :element-type :default now allow character and binary
997 (unsigned-byte 8) I/O
998 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
999 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1001 * bug fix: no more highly sporadic "couldn't check whether ~S is
1002 readable" when reading a stream and an interrupt hits in the middle
1003 of a select system call
1004 * compiler better recognizes complex arrays (reported by Eduardo
1006 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1007 for variables on SBCL built with linkage-tables. (reported by Luis
1009 * various error reporting improvements.
1010 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1011 (Thanks to Thiemo Seufer)
1012 * bug fix: nested reader invokations work correctly
1013 * bug fix: it is possible to have simultaneous references to foreign
1014 code and foreign data with the same name.
1016 ** added x86-64 support
1017 ** incompatible change: the threading api now works with thread
1018 objects instead of thread ids
1019 ** bug fix: threads are protected from signals and interruption when
1020 starting up or going down
1021 ** bug fix: a race where an exiting thread could lose its stack to gc
1022 ** bug fix: don't halt on infinite error in threads if possible
1023 ** fixed numerous gc deadlocks introduced in the pthread merge
1024 ** bug fix: fixed thread safety issues in read and print
1025 ** bug fix: debugger doesn't hang on session lock if interrupted at
1026 an inappropriate moment
1027 ** bug fix: run-program is now thread safe(r)
1028 ** bug fix: inner with-recursive-lock no longer releases the mutex
1029 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1030 * fixed some bugs revealed by Paul Dietz' test suite:
1031 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1032 EXPECTED-TYPE slots filled.
1033 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1035 ** MAKE-CONDITION accepts classes as type-designators.
1036 ** COMPILE may never return NIL.
1037 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1038 range before calling Unix time functions
1040 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1041 * numerous signal handling fixes to increase stability
1042 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1043 * minor incompatible change: we now correctly canonize default
1044 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1045 as per the MOP, rather than the historical (INITARG INITFUNCTION
1046 INITFORM). (reported by Bruno Haible)
1047 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1048 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1049 for more information.
1050 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1051 pathname is a directory pathname.
1052 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1053 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1055 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1056 :METACLASS argument in addition to a class name. (reported by
1057 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1058 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1059 (reported by Bruno Haible; patch from Gerd Moellmann)
1060 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1062 * bug fix: more cleanups to the floating point exception handling on
1063 x86-64 (thanks to James Knight)
1064 * bug fix: the compiler does not try to constant fold calls of
1065 COERCE and BIT-* functions when they return freshly constructed
1066 objects. (reported by Nikodemus Siivola and Paul Dietz)
1067 * optimization: improved the MIPS versions of generic (in the
1068 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1069 * optimization: direct conversion of (unsigned-byte 32) to floats on
1070 the PowerPC platform.
1071 * optimization: structure instances with raw slots now use less
1072 memory, and probably show better memory locality. (thanks to
1074 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1075 are treated as requests for stack allocation on the x86-64,
1076 Alpha-32, PPC and SPARC platforms.
1077 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1078 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1079 (thanks to Tony Martinez)
1081 ** gcing a dead thread can no longer lead to lockups
1082 ** threads block signals until they are set up properly
1083 ** errno is no longer shared by threads
1084 ** interrupt-thread restores the eflags register on x86
1085 ** fixed some lockups due to gc/thread interaction
1086 * fixed some bugs revealed by Paul Dietz' test suite:
1087 ** invalid dotted lists no longer raise a read error when
1088 *READ-SUPPRESS* is T
1089 ** signal an error if a symbol that names a declaration is used
1090 as the name of a type, or vice versa
1091 ** allow using the (declare (typespec var*)) abbreviation for
1092 (declare (type typespec var*)) with all type specifiers
1093 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1094 ** always return NIL from PPRINT-POP when OBJECT is NIL
1095 ** don't signal errors when pretty-printing malformed LABELS,
1096 FLET or MACROLET forms
1097 ** declarations in a DOLIST body are also in scope for the
1099 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1101 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1104 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1105 * fixed cross-compiler leakages that prevented building a 32-bit
1106 target with a 64-bit host compiler.
1107 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1108 opened with :IF-EXISTS :OVERWRITE.
1109 * fixed bug 281: error for an invalid qualifier in a short-form method
1110 combination method is not signalled until the faulty method is called.
1111 * bug fix: iteration variable type inferrer failed to deal with open
1112 intervals. (reported by Alan Shields)
1113 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1114 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1115 or more runtime options were provided to the sbcl binary.
1116 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1117 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1118 * optimization: there's now a fast-path for fixnum arguments in the
1119 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1120 * optimization: the code generated on x86-64 is more compact thanks
1121 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1122 * bug fix: floating-point exception handling now partly works on
1123 x86-64. (Thanks to James Knight)
1124 * improvement to the MIPS backend from Thiemo Seufer: C-style
1125 64-bit long long arguments and return values to alien functions
1127 * fixed some bugs revealed by Paul Dietz' test suite:
1128 ** the type-error signalled from WARN has a filled-in DATUM slot.
1129 ** the type-error required when a stream is not associated with
1130 a file has the stream as its datum.
1131 ** type-errors on single-floats on x86-64 no longer have
1132 :INVALID-OBJECT as the datum
1133 ** the type-errors signalled for invalid function names now have
1134 a correct expected type
1135 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1136 for structures defined with a :TYPE. Documentation strings for
1137 typed structures are no longer immediately discarded
1138 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1139 broadcast streams and synonym streams. FILE-LENGTH now also works
1140 on broadcast streams.
1142 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1143 * incompatible change: the --noprogrammer option, deprecated since
1144 version 0.7.5, has been removed. Please use the equivalent
1145 --disable-debugger option instead.
1146 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1148 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1149 inoperational for a while now) has been completely removed.
1150 * null lexical environments are now printed as #<NULL-LEXENV>,
1151 significantly reducing the amount of clutter in typical backtraces.
1152 * documentation on weak pointers, finalization, and after GC hooks
1153 has been added to the manual.
1154 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1155 as other specialized array types, is much faster. SUBSEQ and
1156 COPY-SEQ on such arrays have also been sped up.
1157 * optimization: EQL is now more efficient when at least other argument
1158 is known to be of type (OR FIXNUM (NOT NUMBER)).
1159 * fixed bug: the runtime is now less vulnerable to changes in the
1160 size of the SBCL object on OS X, and virtual memory is reserved for
1161 all spaces that need to be at a fixed address.
1162 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1163 * fixed bug: finalizers and after GC hooks that cause consing are now
1165 * fixed bug: compiler error messages and summaries are now printed to
1166 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1167 * fixed inference of the upper bound of an iteration variable.
1168 (reported by Rajat Datta).
1169 * fixed bug 211e: calling local functions with duplicated constant
1170 keyword argument no longer causes a bogus style warning about an
1172 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1173 ftype declaration to be disregarded. (reported by Dave Roberts)
1174 * fixed bug 373: caused by erronous compilation of references to alien
1175 variables in the runtime on ppc/darwin.
1176 * fixed bug 376: CONJUGATE type deriver.
1177 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1178 (reported by Baughn on #lisp)
1179 * a cross-compiler bug on non-x86 platforms has been identified and
1180 fixed. (thanks to Bruno Haible)
1181 * improvements to the MIPS runtime code for increased stability.
1182 (thanks to Thiemo Seufer)
1183 * increased the maximimum compact environment size to allow
1184 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1185 * partial workaround for bug 135: don't name the function we're
1186 calling for hairy cases of EVAL, so as not to accumulate one environment
1187 entry per call to EVAL. (reported by Kevin Reid)
1188 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1189 * optimization: type testing for non-vector arrays should be faster.
1190 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1192 * bug fix: setting 31st element of a bit vector to zero did not work
1194 * bug fix: redefining a class definition which failed due to a
1195 previous accessor / function clash now works (but see BUGS entry
1196 #380 for more problems in this area). (thanks to Zach Beane)
1197 * the long form of DEFINE-METHOD-COMBINATION disables method group
1198 checking when given a single method group with pattern *. (thanks
1200 * on x86 compiler supports stack allocation of results of simple
1201 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1202 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1203 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1204 * fixed some bugs related to Unicode integration:
1205 ** the restarts for recovering from input and output encoding
1206 errors only appear when there is in fact such an error to
1208 * fixed some bugs revealed by Paul Dietz' test suite:
1209 ** MISC.549 and similar: late transformation of unsafe type
1210 assertions into derived types caused unexpected code
1212 ** SCALE-FLOAT type deriver is less wrong.
1213 ** type derivers for EXP, LOG and similar functions compute result
1214 types for complex arguments better.
1215 ** (MISC.563) CONJUGATE type deriver works for very restricted
1217 ** out-of-line type testers for character strings are available.
1218 ** EQUAL compiler transform understands specialness of objects
1220 ** accessing double-floats stored on the stack now works on x86-64.
1221 ** debugger internals could sometimes create invalid lispobjs,
1222 resulting in GC crashes.
1223 ** MISC.548: type check weakening can convert required type into
1225 ** initialization forms for bindings are not in scope of free special
1228 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1229 * incompatible change: thread support for non-NPTL systems has
1230 been removed - locking is buggy and unreliable. A threaded
1231 SBCL build will now warn at startup and refuse to create
1232 new threads, unless futex support is detected in the kernel
1233 * incompatible change: the top level REPL now has only an ABORT
1234 restart associated with it, not TOPLEVEL and ABORT as it used to.
1235 TOP and TOPLEVEL are now available as debugger commands for
1236 returning to the top level.
1237 * incompatible change: forms evaluated in the REPL now use the
1238 global optimization policy.
1239 * incompatible change: user- and system-initialization files are
1240 no longer processed with LOAD, but by READ and EVAL; hence the
1241 global optimization policy, startup package, readtable, etc,
1243 * The .fasl file format number has been incremented because of
1244 various incompatible changes.
1245 * internal entry point details and argument counts no longer appear
1246 in backtraces unless explicitly requested by setting
1247 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1248 * built-in and standard functions no longer have names like "top
1249 level local call to FOO".
1250 * fixed bug 32: functions defined in non-null lexical environments
1251 now have more legible printed representation
1252 * fixed bug 33: functions defined in non-null lexical environemnts
1253 are now more amenable to inspection by INSPECT.
1254 * workaround for bug 354: XEPs no longer appear in backtraces unless
1255 explicitly requested.
1256 * fixed bug: receiving the signal which results from attempting to
1257 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1258 SIGBUS on other *BSDs) on architectures where the C stack is also
1259 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1260 to debugger confusion or wild instability. (reported by Cheuksan
1262 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1263 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1264 (reported by Lutz Euler)
1265 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1266 compiled code. (reported by Paul Dietz)
1267 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1268 specialized parameter to a method to something that is not TYPEP
1269 the specializer is now possible.
1270 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1271 face of package deletion.
1272 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1273 pathname. (reported by David Tolpin on c.l.l)
1274 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1275 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1276 STRING-STREAM and FILE-STREAM, respectively.
1277 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1278 than previously; in addition, its test suite is now run on build.
1279 (thanks to Luke Gorrie)
1280 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1281 * optimization: added a immediate representation for single-floats
1283 * fixed some bugs revealed by Paul Dietz' test suite:
1284 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1285 ** attempting to create a package with a colliding nickname causes
1286 correctable errors to be signalled.
1287 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1288 ** several x86-64 backend bugs related to sign-extension of immediate
1291 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1292 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1293 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1295 * fixed loading of multiply forward-referenced layouts.
1296 (thanks to Cheuksan Wang)
1297 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1298 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1299 (see COMPILE-FILE documentation for details of :PRINT option)
1300 * fixed bugs 19 and 317: fixed-format floating point printing is
1301 more accurate. This also fixes a bug reported by Adam Warner
1302 related to the ~@F format directive.
1303 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1305 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1306 dispatch table if the from-char is a dispatch macro character.
1307 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1308 by the wordsize no longer ignore the last word. (reported by Lutz
1310 * fixed bug in type checking of dynamic-extent variables. (reported
1312 * optimization: sequence traversal functions use their freedom to
1313 coerce function designators to functions.
1314 * optimization: code with many calls to CLOS methods specialized on
1315 CLOS classes has had redundant type checks removed.
1316 * optimization: added declarations to speed up operations that access
1317 the internal character database (for example STRING-UPCASE)
1318 * optimization: comparison operations between floats and sufficiently small
1319 fixnums no longer create extra rationals
1320 * fixed some bugs related to Unicode integration:
1321 ** portions of multibyte characters at the end of buffers for
1322 character-based file input are correctly transferred to the
1323 start of the buffer at the next read.
1324 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1325 passing it through to OPEN.
1326 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1327 argument given to internal calls to OPEN.
1328 ** the built-in comment readers (introduced by character sequences
1329 ";" and "#|") are more forgiving to encoding errors; they will
1330 STYLE-WARN and then attempt to resync the stream at a character
1331 boundary. (thanks to Teemu Kalvas)
1332 * fixed some bugs in the x86-64 port:
1333 ** Negative short int return values from c-calls are sign-extended
1335 ** The stack is aligned to 16-bytes for c-calls, as required by
1336 the ABI. (thanks to Cheuksan Wang)
1337 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1338 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1339 for enums. (thanks to Vincent Arkesteijn)
1340 ** Multiple small optimizations and bugfixes for floating point
1342 * fixed some bugs revealed by Paul Dietz' test suite:
1343 ** Space, Tab, Linefeed, Return and Page have the invalid
1344 secondary constituent character trait.
1345 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1347 ** WITH-INPUT-FROM-STRING should only update the index place on
1349 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1350 works more reliably.
1351 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1352 with one-dimensional arrays with fill pointers.
1353 ** TYPE-OF failed on a complex with an integer realpart and a
1355 ** compiler failure during type inference for the code of form
1356 (IF (EQL X (THE ...)) ...) (MISC.535).
1358 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1359 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1360 platform. The port supports 61-bit fixnums, large memory spaces
1361 and reloading shared object files.
1362 * enhancement: saving cores with foreign code loaded is now
1363 supported on ppc/Darwin in addition to the previously supported
1365 * enhancement: the statistical profiler now walks deeper into the
1366 call stack for more meaningful call-graphs and accrued time
1367 reports (x86/x86-64 only). It also now reports time spent in
1369 * enhancement: it is now possible to trace most individual methods
1370 of a generic function in addition to tracing the generic function
1372 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1374 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1375 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1376 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1377 produces an error. (thanks to Vincent Arkesteijn)
1378 * bug fix: NAMESTRING on pathnames with :WILD components in their
1379 directories works correctly. (thanks to Artem V. Andreev)
1380 * fixed bug 125: compiler preserves identity of closures. (reported
1382 * bug fixed: functions with &REST arguments sometimes failed with
1383 "Undefined function" when compiled with (DEBUG 3). (reported by
1385 * bug fix: overflow during compiling of setting element of a bit
1386 vector with constant index and value. (reported by Timmy Douglas)
1387 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1388 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1389 * on x86 compiler partially supports stack allocation of dynamic-extent
1391 * GO and RETURN-FROM do not check the extent of their exit points
1392 when compiled with SAFETY 0.
1393 * fixed some bugs related to Unicode integration:
1394 ** encoding and decoding errors are now much more robustly
1395 handled; it should now be possible to recover even from invalid
1396 input or output to the terminal. (thanks to Teemu Kalvas)
1397 ** provided a first cut at implementing STRING-TO-OCTETS and
1398 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1399 ** altered the SB-MD5 contributed module slightly, changing the
1400 interface just enough for it to be supportable for builds where
1401 lisp characters are not eight bits.
1402 * fixed some bugs revealed by Paul Dietz' test suite:
1403 ** the FORMATTER-generated functions for ~V[ conditionals require
1404 the correct number of arguments.
1405 ** READ-FROM-STRING returns the mandated second value when applied
1406 to displaced strings.
1407 ** the #\Rubout and #\Backspace characters are treated as invalid
1408 constituent characters by the tokenizer.
1410 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1411 * new feature: reloading changed shared object files with
1412 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1413 * new feature: references to foreign variables and functions
1414 can now be compiled and loaded before the corresponding shared
1415 object file is loaded, as long as the foreign definitions are
1416 available at runtime.
1417 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1418 supported. (thanks to Dan Debertin)
1419 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1420 just keywords. (thanks to Vincent Arkesteijn)
1421 * fixed bug #331: structure-class instances corresponding to
1422 DEFSTRUCT forms are now created eagerly.
1423 * fixed bug #345: backtraces from calls to undefined functions work
1424 on x86 as well. Related bug #61 is now also partially fixed on x86
1425 (backtraces from throws to unknown catch tags.)
1426 * bug fix: lambda-list parsing is now stricter vrt. order and number
1427 of lambda-list keywords.
1428 * bug fix: as specified by AMOP, an error is signalled if a
1429 class-option appears multiple times in a DEFCLASS form. (reported
1431 * bug fix: RANDOM can be compiled when the compiler derives the type
1432 of its numeric argument as a disjoint set of small integers.
1433 (reported by Paul Dietz)
1434 * bug fix: starting a core saved with shared objects loaded when
1435 those objects are not available no longer causes threaded SBCL to
1436 hang. (reported by Sean Ross)
1437 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1439 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1440 argument, which must always be NIL. (reported by Kalle Niemitalo)
1441 * bug fix: printing 1.0d+23 no longer results in an error.
1442 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1443 * bug fix: structure slot setters preserve evaluation order. (thanks
1445 * fixed some bugs related to Unicode integration:
1446 ** RUN-PROGRAM can allow its child to take input from a Lisp
1447 stream. (reported by Stefan Scholl)
1448 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1449 locales. (reported by Ken Causey)
1450 * fixed some bugs revealed by Paul Dietz' test suite:
1451 ** INCF, DECF and REMF evaluate their place form as specified in
1453 ** FORMATTER expands ~{ ~} iteration directives with V or #
1454 parameters correctly.
1455 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1456 consequent uses no arguments correctly.
1457 ** the system has a partial understanding of the (COMPLEX RATIO)
1459 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1462 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1463 * new feature: a build-time option (controlled by the :SB-UNICODE
1464 keyword feature, enabled by default) for building the system with
1465 support for the entire 21-bit character space defined by the
1467 * new feature: the system now has rudimentary external-format
1468 support; the primary user-visible change at this time is that
1469 characters with the high bit set (such as lower-case-e-acute) will
1470 print correctly to a terminal in a UTF-8 environment.
1471 * minor incompatible change: BASE-CHAR no longer names a class;
1472 however, CHARACTER continues to do so, as required by ANSI.
1473 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1474 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1475 has been moved to the SB-EXT package (temporarily re-exported from
1477 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1478 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1480 * fixed bug #351: better error-handlind and reporting for malformed
1482 * fixed bug #350: bignum-printing is now more memory-efficient,
1483 allowing printing of very large bignums, eg. (expt 2 10000000).
1484 (reported by Bruno Haible)
1485 * fixed bug #302: better primitive-type selection for intersection
1487 * fixed bug #308: non-graphic characters now all have names, as
1488 required. (reported by Bruno Haible)
1489 * bug fix: redefining a class with different superclasses now correctly
1490 removes it from the direct-subclasses of its previous superclasses.
1491 (reported by David Morse)
1492 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1493 new value now works. (reported by Bruno Haible)
1494 * bug fix: correct canonicalization of multiple non-standard slot
1495 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1496 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1497 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1498 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1499 now exists, an signals an error.
1500 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1501 during macroexpansion for non-iteration variables as well. (reported
1502 by Bruno Haible for CMUCL)
1503 * bug fix: Cyclic structures and unprintable objects in compiler
1504 messages no longer cause errors. (reported by Bruno Haible)
1505 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1506 and READ-FROM-STRING all now return a primary value of NIL if
1507 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1508 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1509 (reported by Bruno Haible for CMUCL)
1510 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1511 do not designate a condition. (reported by Bruno Haible for
1513 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1514 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1515 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1516 permitting resolution in favour of any of the conflicting symbols.
1517 (reported by Bruno Haible for CMUCL)
1518 * bug fix: EQUAL compiler optimizations is less aggressive on
1519 strings which can potentially compare true despite having distinct
1520 specialized array element types.
1521 * bug fix: unit enumerations can be defined without dividing by
1522 zero. (thanks to Vincent Arkesteijn)
1523 * FORMAT compile-time argument count checking has been enhanced.
1524 (report from Bruno Haible for CMUCL)
1525 * a partial workaround for the bug 262: the compiler does not try to
1526 inline-expand a local function doing RETURN-FROM from a deleted
1527 BLOCK. (thanks to Peter Denno for the bug report and to David
1528 Wragg for the simple test case)
1529 * fixed some bugs revealed by Paul Dietz' test suite:
1530 ** RENAME-PACKAGE allows all package designators as new package
1532 ** constraint propagation and lambda variable substitution are
1533 more cautious in dealing with partially deleted code.
1534 ** compiler sometimes lost reoptimization passes.
1535 ** CERROR, given a condition as condition designator, passes its
1536 remaining arguments to the continue format control without
1538 ** Case-altering FORMAT directives work correctly on non-ASCII
1540 ** The REAL type specifier handles bounds outside the floating
1541 point ranges without signalling FLOATING-POINT-OVERFLOW.
1542 ** Functions with IR1-transformations can create intercomponent
1543 references to global functions.
1544 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1546 ** FORMAT ~R treats a nil value for its first parameter correctly.
1548 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1549 * enhancement: saving cores with foreign code loaded is now
1550 supported on x86/NetBSD and sparc/Linux in addition to the previously
1551 supported platforms.
1552 * bug fix: on some platforms repeated installations caused multiple
1553 copies of HTML documentation to be installed -- should not happen
1554 any more. (reported by Stefan Scholl)
1555 * bug fix: parsing self-recursive alien record types multiple times
1556 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1557 original patch by Helmut Eller for CMUCL)
1558 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1559 (thanks to Richard Kreuter)
1560 * bug fix: defining classes whose accessors are methods on existing
1561 generic functions in other (locked) packages no longer signals
1562 bogus package lock violations. (reported by François-René Rideau)
1563 * bug fix: special variables as DEFMETHOD parameters no longer have
1564 associated bogus type declarations. (reported by David Wragg and
1566 * bug fix: read-write consistency on streams of element-type
1567 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1568 * bug fix: redefiniton of the only method of a generic function with
1569 no DEFGENERIC no longer emits a full WARNING. In addition,
1570 redefinition of generic functions with no DEFGENERIC to an
1571 incompatible lambda list now signals an error. (thanks to Zach
1573 * bug fix: DEFGENERIC now works even when there's a function of the
1574 same name in an enclosing lexical environment. (thanks to Zach
1576 * fixed compiler failure, caused by instrumenting code during
1577 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1578 * optimization: added loop analysis and improved register allocation
1579 * fixed some bugs revealed by Paul Dietz' test suite:
1580 ** POSITION on displaced vectors with non-zero displacement
1581 returns the right answer.
1582 ** (SIMPLE-STRING) is a valid type specifier for sequence
1584 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1586 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1587 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1589 ** The justification version of the FORMAT ~< directive treats
1590 non-zero minpad parameter correctly.
1592 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1593 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1594 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1595 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1596 the supported interface.
1597 * new feature: Single-stepping of code compiled with DEBUG 2 or
1598 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1599 * new feature: saving cores with foreign code loaded is now
1600 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1601 Timothy Moore's work for CMUCL)
1602 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1603 parameters to * instead of NIL if no initform is supplied.
1604 (reported by Johan Bockgaard)
1605 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1606 correctly when FUNCALL appears as the car of the form. Note:
1607 despite this FUNCALL forms are not currently subject to
1608 compiler-macro expansion. (port of Raymond Toy's fix for the
1609 same from CMUCL, reported by Johan Bockgaard)
1610 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1611 (thanks for Teemu Kalvas)
1612 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1613 case instead of terminating the iteration (thanks for Julian
1614 Squires, Sean Champ and Raymond Toy)
1615 * bug fix: incorrect expansion of defgeneric that caused a style
1616 warning. (thanks for Zach Beane)
1617 * bug fix: slot accessor effective method computation works properly
1618 for classes with multiple non-standard applicable methods on
1619 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1620 * on x86 compiler supports stack allocation of results of LIST and
1621 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1622 CMUCL implementation by Gerd Moellmann)
1623 * fixed some bugs revealed by Paul Dietz' test suite:
1624 ** FORMAT strings with both the ~<~:;~> form of the justification
1625 directive and pretty-printing directives cause an error.
1627 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1628 * incompatible change: the internal functions
1629 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1630 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1631 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1632 instead of the old functions.
1633 * new feature: on platforms where "dladdr" is available foreign
1634 function names now appear in backtraces. (based on Helmut Eller's
1636 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1637 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1639 * documentation: ASDF manual is now distributed with SBCL.
1640 * bug fix: SBCL can now load its contributed modules with REQUIRE
1641 even if the system-provided entries have been removed from
1642 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1644 * bug fix: backtraces involving undefined functions or assembly
1645 routines are more informative. (thanks to Brian Downing)
1646 * bug fix: mutually referent alien structures now work correctly.
1647 (reported by Rick Taube)
1648 * bug fix: structures defined by WITH-ALIEN can be referred to
1649 within other definitions in the same WITH-ALIEN.
1650 * bug fix: division operators (MOD, TRUNCATE and the like) with
1651 constant zero divisors and integer dividends no longer generate
1653 * bug fix: provide default methods for INPUT-STREAM-P and
1654 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1655 * bug fix: improve the detection and resolution of MOP metacycles
1656 (where effective method computation depends on the generic function
1657 having its effective method computed). (reported by Bruno Haible)
1658 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1659 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1660 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1661 * bug fix: pathnames with relative directory components can be
1662 represented relative to default pathnames.
1663 * optimization: in taking the GCD of bignums, reduce the two bignums
1664 to approximately the same size (using Euclid's algorithm) before
1665 applying the more sophisticated binary GCD. (thanks to Juho
1667 * optimization: COUNT on bitvectors now operates word-at-a-time.
1668 * optimization: ASH with a positive, but not necessarily constant,
1669 (leftwards) shift, when in a modular context, is compiled to a
1671 * fixed some bugs revealed by Paul Dietz' test suite:
1672 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1673 if the corresponding argument is NIL.
1674 ** FORMAT directives accepting a minpad parameter treat negative
1676 ** literal commas as character parameters to FORMAT directives are
1678 ** literal spaces directly after ~<Newline> directives within a
1679 format-logical-block (~:< ~@:>) do not induce :FILL-style
1680 conditional newlines.
1681 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1682 as specified: it no longer includes conditional newlines).
1683 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1685 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1686 :PER-LINE-PREFIX argument does not evaluate to a string.
1688 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1689 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1690 the manual for details; for now, package locks can be disabled by
1691 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1692 if no major problems are found then it is likely that they will be
1693 compiled in unconditionally.
1694 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1695 unsupported operators on all platforms. To load a shared library
1696 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1697 object file, link it into a shared library outside of SBCL and
1698 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1700 * bug workaround: ROOM T and the GC were not working together
1701 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1702 reporting facility aren't preserved. That reporting has been
1703 disabled, so now until and unless someone figures out how to make it
1704 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1705 * minor incompatible change: as threatened around sbcl-0.8.0, the
1706 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1707 The symbols are also exported from SB-PCL for backwards
1708 compatibility, but more so than before SB-PCL should be treated as
1709 an implementation-internal package.
1710 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1712 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1713 It is known as of this release not to work on the Alpha, however.
1714 * fixed bug #167: errors signalled due to illegal syntax in method
1715 bodies are now more legible.
1716 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1717 designators and can hence be used with TYPEP.
1718 * fixed bug #333: CHECK-TYPE now ensures that the type error
1719 signalled, if any, has the right object to be accessed by
1720 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1721 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1722 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1724 * fixed bug #334: programmatic addition of slots using specialized
1725 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1726 and :ALLOCATION :CLASS slots.
1727 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1728 float exponents. (rereported by Peter Seibel)
1729 * fixed a bug: #\Space (and other whitespace characters) are no
1730 longer considered to be macro characters in standard syntax by
1731 GET-MACRO-CHARACTER.
1732 * fixed bug: initialization of condition class metaobjects no longer
1733 causes an instance of the condition to be created. (reported by
1735 * fixed bug: it is now possible to have slots such that
1736 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1737 description is neither :INSTANCE nor :CLASS.
1738 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1739 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1740 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1741 non-standard SLOT-DEFINITION-ALLOCATION values.
1742 * fixed bugs: various race conditions handling exiting threads.
1743 CL-PPCRE's thread torture test now passes.
1744 * fixed arguably-a-bug: GC time proportional to number of threads in
1745 system even when most of them are idle
1746 * optimization: improved performance of BIT and SBIT on bit-vectors.
1747 * .fasl file incompatibility: The fasl file version number has
1748 been incremented because of changes associated with package locks.
1750 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1751 * minor incompatible change: the system no longer provides
1752 optimization hints (conditions which are TYPEP
1753 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1754 mode; these hints are still emitted when higher SPEED optimization
1756 * new contrib module: a sampling profiler (profiling by statistical
1757 sampling, rather than by instrumenting functions) is available as
1758 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1759 * the behaviour of the standard function ED is now customizeable by
1760 third parties through a hook variable: see ED's documentation
1761 string for information on the protocol.
1762 * the compiler no longer emits efficiency notes for (FUNCALL X)
1763 when the type of X is uncertain under default optimization
1765 * fixed bug 276: mutating a binding of a specialized parameter to a
1766 method to something that is not TYPEP the specializer is now
1768 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1769 exhibit the required behaviour on the x86 platform. (thanks to
1770 Peter van Eynde, Eric Marsden and Bruno Haible)
1771 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1772 even for difficult arguments. (reported by Peter Graves)
1773 * fixed bug 141a: the backquote printer now descends quoted
1775 * fixed another bug in backquote printing: no more destructive
1776 modification of the form's list structure. (reported by Brian
1778 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1779 move between its address being taken and the call to
1780 interrupt_thread, fixing a crashing race condition.
1781 * the SB-POSIX contrib implementation has been adjusted so that it
1782 no longer exhibits ridiculously poor performance when constructing
1783 instances corresponding to C structs.
1785 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1786 * minor incompatible change: the sb-grovel contrib now treats C
1787 structures as alien (in the sense of SB-ALIEN) objects rather than
1788 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1789 has implications for memory management of client code
1790 (sb-grovel-returned objects must now be manually managed) and for
1791 type safety (alien objects now have full types).
1792 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1793 used to control emission of compiler diagnostics, rather than the
1794 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1795 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1796 quality should be considered deprecated.
1797 * install.sh now installs the user manual as well
1798 * (not quite a new documentable feature, but worth considering in
1799 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1800 beginnings of a semantically meaningful condition hierarchy is
1801 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1803 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1804 designator as the defaults argument.
1805 * fixed bug: Displaced arrays whose displaced-to array has become
1806 too small now cause ARRAY-DIMENSION to signal an error, providing
1807 for safer bounds-checking. (reported by Bruno Haible)
1808 * fixed bug: DEFCLASS slot definitions with identical :READER and
1809 :WRITER names now signal a reasonable error. (reported by Thomas
1811 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1813 * fixed bug: Invalid slot specification errors now print correctly.
1814 (thanks to Zach Beane)
1815 * fixed bug 320: Shared to local slot value transfers in class
1816 redefinitions now happen corrently from superclasses as
1817 well. (reported by Bruno Haible)
1818 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1820 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1821 improper lists correctly. (reported by Bruno Haible)
1822 * fixed bug 313: source-transform for <fun-name> was erroneously
1823 applied to a call of a value of a variable with name <fun-name>.
1824 (reported by Antonio Menezes Leitao)
1825 * fixed bug 307: The obsolete instance protocol ensures that
1826 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1827 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1828 remove dead unknown-values globs from the middle of the stack.
1829 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1830 easier to resume long computations after using *BREAK-ON-SIGNALS*
1831 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1832 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1833 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1834 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1835 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1836 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1838 * fixed a (fairly theoretical) bug in string printing: if
1839 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1840 does not have array-element-type equal to the most general string
1842 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1843 function name. (thanks to Nikodemus Siivola)
1844 * fixed bug: compile-time format string checker failed on
1845 non-closed ~{. (reported by Thomas F Burdick)
1846 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1847 don't warn on legal loop code involving more than one aggregate
1848 boolean. (thanks to Nikodemus Siivola)
1849 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1850 in keyword position to MAKE-PACKAGE was not being reported as
1852 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1853 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1855 * fixed bugs 280 and 312: the checking for multiple definitions in a
1856 file is less likely to become confused by uses of inline
1858 * fixed bug: the #S reader macro performs the keyword coercion
1859 specified for slot names. (reported by Kalle Niemitalo)
1860 * fixed bug: lambda lists may contain symbols whose names start with
1861 & but are not lambda-list-keywords; their occurrence triggers a
1863 * fixed bug 321: define-method-combination argument lambda lists do
1864 not cause invalid code to be generated when &optional or &aux
1865 variables have default values. (reported by Bruno Haible)
1866 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1867 have CLOS classes; this appears also to have fixed reported
1868 instability in stack exhaustion detection.
1869 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1870 MUFFLE-WARNING when no associated restart is present is now
1872 * optimization: rearranged the expansion of various defining macros
1873 so that each expands into only one top-level form in a
1874 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1876 * optimization: used a previously little-used slot in symbols to
1877 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1879 * fixed some bugs revealed by Paul Dietz' test suite:
1880 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1882 ** Fixed an optimization bug related to inheritance of initforms
1883 from local to shared slots.
1884 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1885 ** CAST optimizer forgot to flush argument derived type.
1886 ** print/read consistency on floats is now orders of magnitude
1887 more likely. (thanks also to Bruno Haible for a similar report
1889 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1890 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1891 for objects of type REAL. Make it so.
1892 ** SXHASH is specified (infelicitously) to respect similarity,
1893 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1894 it so. (thanks to Markus Ziegler)
1895 ** on the Alpha, the compiler succeeds in compiling functions
1896 returning a known number of arguments greater than 63.
1897 ** fixed handling of invalid NIL arguments in keyword position
1899 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1901 ** provided a readably-printable representation for RANDOM-STATE
1903 ** ensured that pathnames, where they have a namestring, always
1904 print using #P"..." syntax.
1906 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1907 * Support for the forthcoming 2.0 version of the NetBSD kernel
1908 running on x86 hardware has been added. (thanks to Perry
1909 E. Metzger most immediately, and others for their past work)
1910 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1911 current release version; SBCL's OpenBSD support had been broken
1912 since about the time of OpenBSD's switch to ELF binary format.)
1913 * [placeholder for DX summary]
1914 ** user code with &REST lists declared dynamic-extent, under high
1915 speed or space and low safety and debug optimization policy.
1916 * The manual has been converted to Texinfo format and the debugger
1917 chapter from the cmucl manual has been added.
1918 * A facility has been added to extract documentation strings from
1919 sbcl and store them as Texinfo-formatted snippets for inclusion in
1920 the manual (via Texinfo's @include directive)
1921 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1922 the test case to Dave Roberts)
1923 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1924 pointers. (reported by Sean Ross)
1925 * bug fix: PROFILE output is printed nicely even for large numerical
1926 values. (thanks to Zach Beane)
1927 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1928 greater than 32 handle EOF correctly.
1929 * bug fix: on X86 an immediate argument of the IMUL instruction is
1930 correctly printed in disassembly. (thanks to Lutz Euler)
1931 * bug fix: class slots in redefined classes preserve their old
1932 values. (thanks to Bruno Haible and Nikodemus Siivola)
1933 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1934 succeeds. (reported by Marco Baringer)
1935 * fixed some bugs revealed by Paul Dietz' test suite:
1936 ** READ-SEQUENCE now works on ECHO-STREAMs.
1937 ** RATIONALIZE works more according to its specification. (thanks
1940 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1941 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1942 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1943 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1944 should matter to you only if you rebind the printer control
1945 variables and then find you want different bindings in the
1946 debugger than in the ordinary execution of your program.)
1947 * The runtime build system has been tweaked to support building
1948 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1949 assembler and linker. (thanks to Nikodemus Siivola)
1950 * Unbound, undefined, undeclared variables now trigger full
1951 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1952 more often programmer error than deliberate exploitation of undefined
1954 * optimization: the hash algorithm for strings has changed to one
1955 that is less vulnerable to spurious collisions. (thanks to Juho
1957 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1958 needless bounds checking. (thanks to Juho Snellman)
1959 * optimization: implemented multiplication as a modular
1960 (UNSIGNED-BYTE 32) operation on the PPC backend.
1961 * fixed some bugs revealed by Paul Dietz' test suite:
1962 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1964 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1965 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1966 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1968 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1970 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1972 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1973 their output stream on EOF from read.
1974 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1975 have been read to end-of-file.
1976 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1978 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1979 description of determination of which consecutive characters
1981 ** Printing the "Space" character with escaping on now yields "#\\ ",
1982 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1983 ** Reading floating-point numbers with *READ-BASE* set to a number
1984 less than 10 works correctly.
1985 ** Reading floating-point numbers with *READ-BASE* set to a number
1986 more than 10 works correctly.
1987 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1988 the readtable currently in effect.
1990 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1991 * minor incompatible change: parsing of namestrings on a physical
1992 (Unix) host has changed; numbers after the final #\. in a
1993 namestring are no longer interpreted as a version field. This is
1994 intented to be largely invisible to the user, except that the
1995 meaning of the namestring "*.*.*" has changed: it now refers to a
1996 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1997 should usually be replaced by
1998 (make-pathname :name :wild :type :wild :version :wild)
1999 with the added benefit that this is more likely to be portable.
2000 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2001 option to OPEN now signals an error if the file being opened
2002 exists; this may have an impact on existing code.
2003 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2004 signals received, on the PowerPC platforms (both Linux and
2005 Darwin). (thanks to Pierre Mai for pointing out the location of
2007 * several fixes on OS X: The system now builds and runs cleanly on
2008 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2009 (thanks to Brian Mastenbrook)
2010 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2011 second-resolution: integer multiples of 1/3600 between -24 and 24.
2012 (thanks to Vincent Arkesteijn)
2013 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2014 type when called with 1 argument; PEEK-CHAR checked type of
2015 PEEK-TYPE only after having read first character from a
2016 stream. (reported by Peter Graves)
2017 * bug fix: the garbage collector now has much better locality
2018 behaviour, and in particular no longer treats all memory as being
2019 exhausted when many small objects point to each other in a deeply
2021 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2022 recognized as being TYPEP their class.
2023 * bug fix: the PUSHNEW documentation string has been corrected.
2024 (thanks to Vincent Arkesteijn)
2025 * bug fix: defaulting of the value for the last of an atypically
2026 large number of multiple values being bound was not being
2027 performed correctly on the Alpha or PPC platforms
2028 * optimization: implemented multiplication as a modular
2029 (UNSIGNED-BYTE 32) operation on the x86 backend.
2030 * optimization: SEARCH on simple-base-strings can now be open-coded.
2031 (see also contrib/compiler-extras.lisp for inspiration for
2032 teaching the compiler about the Boyer-Moore algorithm).
2033 * value, returned by MAX (and MIN) called with several EQUALP, but
2034 not EQL, arguments now does not depend on compiler settings.
2035 * fixed some bugs revealed by Paul Dietz' test suite:
2036 ** in stack analysis liveness information is propagated from
2037 non-local entry points.
2038 ** pathwise CAST removing failed when the CAST node did not start
2040 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2041 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2043 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2044 host is already defined.
2045 ** RENAME-FILE works on streams instead of signalling an internal
2047 ** PEEK-CHAR uses the current readtable when determining whether
2048 or not a character is whitespace.
2049 ** MERGE-PATHNAMES handles the case when the pathname does not
2050 specify a name while the default-pathname specifies a version.
2051 ** Pathnames now stand a better chance of respecting print/read
2053 ** Attempting to use standardized file system operators with a
2054 pathname with invalid :DIRECTORY components signals a
2056 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2058 ** DIRECTORY on logical pathnames is more correct.
2059 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2060 signal a TYPE-ERROR if their argument is not a stream.
2061 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2062 designator argument does not designate a stream.
2063 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2064 examining the synonym.
2065 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2067 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2068 element-type larger than ([UN]SIGNED-BYTE 32).
2070 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2071 * When built with the :SB-FUTEX feature, threaded builds now take
2072 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2073 for faster/more reliable mutex and condition variable support.
2074 * Incompatible change (but one you probably shouldn't have been using
2075 anyway): the interface and code for arbitrating between multiple
2076 threads in the same user session has been redesigned.
2077 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2078 error when called without an explicit environment argument.
2079 (thanks to Nikodemus Siivola)
2080 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2081 confused about their position. (thanks to Adam Warner and Gerd
2083 * bug fix: Pretty printing backquoted forms with unquotations in the
2084 argument list position of various code constructs such as LAMBDA
2085 now works correctly. (reported by Paul Dietz)
2086 * bug fix: Pretty printing unquotations no longer loses all
2087 stream position information.
2088 * optimization: performance of string output streams is now less
2089 poor for multiple small sequence writes.
2090 * optimization: performance of CSUBTYPEP in the presence of complex
2091 expressions involving CONS and NOT many times has been improved.
2092 (reported by Paul Dietz)
2093 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2095 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2096 streams for the REPL, for the debugger, and so on. (thanks to
2098 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2099 * fixed some bugs revealed by Paul Dietz' test suite:
2100 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2101 is ignored if it is too small, rather than propagating through
2102 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2103 ** extremely complex negations of CONS types were not being
2104 sufficiently canonicalized, leading to inconsistencies in
2106 ** VALUES tranformer lost derived type.
2108 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2109 * fixed a bootstrapping bug: the build process no longer assumes
2110 that the various BOOLE-related constants have the same value in
2111 host and target lisps. (noted by Paul Dietz' test suite on an
2112 SBCL binary built from CLISP)
2113 * The system can now be dynamically linked on the MIPS platform,
2114 which enables dynamic loading of foreign code from Lisp. (thanks
2115 to Ralf Baechle for discussions on the MIPS ABI)
2116 * The system now records debugging information for its own source
2117 files in a filesystem-position-independent manner, relative to
2118 the "SYS" logical host.
2119 * fixed a compiler bug: MV-LET convertion did not check references
2120 to the "max args" entry point. (reported by Brian Downing)
2121 * tweaked disassembly notes to be less confident about proclaiming
2122 some instruction as an LRA. (thanks to Brian Downing)
2123 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2124 now each have their own history, command character, and other
2125 characteristics. (thanks to David Lichteblau)
2126 * fixed some bugs revealed by Paul Dietz' test suite:
2127 ** compiler failure in compiling LOGAND expressions including a
2129 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2130 shift greater than 32.
2131 ** FLUSH-DEST did not mark blocks for type check regeneration.
2132 ** HANDLER-CASE failed to accept declarations in handler clauses
2133 in some circumstances.
2135 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2136 * New code in contrib/sb-introspect (still probably not entirely
2137 stable yet) provides some support for smart Lisp development
2138 environments like SLIME.
2139 * The conditions signalled for errors occurring when loading .fasl
2140 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2141 in a way which should help ASDF recover gracefully.
2142 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2143 stuff has been cleaned up. If you code contrib/ stuff, this might
2144 affect you, and you can look at contrib/README, contrib/STANDARDS,
2145 and/or the 0.8.4.27 diff to check.
2146 * In full calls the compiler now does not generate checks for declared
2147 argument types for all arguments.
2148 * various threading fixes
2149 ** and some experimental patches which didn't make it into
2150 the main tree for this release, but which are shipped in
2151 contrib/experimental-thread.patch as a possible fix for some
2152 failures (deadlock, spinning...) in GC-intensive multithreaded
2154 * fixed PPC build problem (source code incompatibility of different
2155 library versions): added offsetof() hackery which attempts to divine
2156 where glibc maintainers put uc_mcontext today
2157 * fixed bug 282: compiler does not trust type assertions while passing
2158 arguments to a full call.
2159 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2160 &OPTIONAL VALUES type parameter.
2161 * fix bug 214: algorithm for noting rejected templates is now more
2162 similar to that of template seletion. (also reported by rydis on
2164 * fixed bug 141b: printing backquoted information readably and prettily
2165 inserts a space where necessary.
2166 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2167 (CHARACTER 10) are now reported as errors, rather than propagated
2168 as unknown types. (reported by piso on #lisp)
2169 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2170 with values NIL and :ERROR. (thanks to Milan Zamazal)
2171 * fixed bug 191c: CLOS now does proper keyword argument checking as
2172 described in CLHS 7.6.5 and 7.6.5.1.
2173 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2174 longer raise an error; further, using a list as a for-as-arithmetic
2175 counter now raises a meaningful error.
2176 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2177 now understood by sequence creation functions such as MAKE-SEQUENCE
2179 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2180 a STRING-INPUT-STREAM.
2181 * compiler enhancement: SIGNUM is now better able to derive the type
2183 * type declarations inside WITH-SLOTS are checked. (reported by
2185 * fixed some bugs revealed by Paul Dietz' test suite:
2186 ** incorrect optimization of TRUNCATE for a positive first
2187 argument and negative second.
2188 ** compiler failure in let-convertion during flushing dead code.
2189 ** compiler failure while deriving type of TRUNCATE on an
2190 interval, containing 0.
2191 ** ASH of a negative bignum by a negative bignum count now returns
2193 ** intersection of CONS types now canonicalizes properly, fixing
2194 inconsistencies in SUBTYPEP.
2196 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2197 * incompatible change: The --disable-debugger command line
2198 option now clobbers the debugger at a more fundamental
2199 level, by redefining #'INVOKE-DEBUGGER instead of by
2200 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2201 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2202 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2203 enter the debugger REPL and then suffer recursive errors
2204 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2205 while under the new system, BREAK in --disable-debugger mode
2206 terminates the system just as an unhandled error would.
2207 * fixed compiler performance when processing loops with a step >1;
2208 * bug fix: DOCUMENTATION now retrieves generic function
2209 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2210 support has been systematized, and now supports the methods
2211 specified by ANSI, along with a default method and a method for
2212 slot documentation. (reported by Nathan Froyd)
2213 * bug fix: effective methods associated with a generic function are
2214 no longer cached over a change of that generic function's method
2215 combination. (reported by Andreas Fuchs)
2216 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2217 names a non-existent directory. (thanks to Andreas Fuchs)
2218 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2219 the PPC platform, lead to stack corruption; this has been fixed.
2220 (reported by Rainer Joswig)
2221 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2222 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2223 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2224 arithmetic in general, is now much more reliable on the PPC
2226 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2227 platform now returns the right answer.
2228 * optimization: restored some effective method precomputation in
2229 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2230 precomputation is now tunable.
2231 * optimization: compiler-internal data structure use has been
2232 reviewed, and changes have been made that should improve the
2233 performance of the compiler by about 20%.
2234 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2235 simple-base-strings and simple-bit-vectors is improved.
2236 * optimization: the optimization of 32-bit logical and arithmetic
2237 functions introduced in version 0.8.3 on the x86 has been
2238 implemented on the mips, ppc and sparc platforms; an
2239 implementation of the same facility, but for 64-bit arithmetic,
2240 has been added for the alpha.
2241 * microoptimization: the compiler is better able to make use of the
2242 x86 LEA instruction for multiplication by constants.
2243 * bug fix: in some situations compiler did not report usage of
2244 generic arithmetic in (SPEED 3) policy.
2245 * bug 145b fix: compiler used wrong type specifier while converting
2246 MEMBER-types to numeric.
2247 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2249 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2250 index argument now works properly on the Alpha platform.
2251 * bug fix: floating point exception treatment on the Alpha platform
2253 * bug fix: FILE-POSITION works much better on string input and
2254 output streams. (thanks to Nikodemus Siivola)
2255 * bug fix: many threading/garbage collection symptoms sorted.
2256 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2257 might be pseudo-atomic.
2258 * internal change: Stopping for GC is now done with signals not
2259 ptrace. GC is now done in whichever thread wanted it, instead of
2261 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2263 * bug fix: result form in DO is not contained in the implicit
2265 * incompatible change: ICR structure is changed; the value part of
2266 CONTINUATION is now called LVAR; corresponding functions are
2267 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2268 * added type deriver for ISQRT (thanks to Robert E. Brown).
2269 * bug fix: better support for loading from the command line when an
2270 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2272 * fixed some bugs revealed by Paul Dietz' test suite:
2273 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2274 ** ROUND and FROUND now give the right answer when given very
2275 small float arguments.
2276 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2278 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2279 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2280 ** LCM should return a non-negative integer.
2281 ** PARSE-INTEGER returned the index of a terminator instead of the
2282 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2283 ** PARSE-INTEGER returned an incorrect index being applied to a
2285 ** LCM with two arguments of 0 returns 0 rather than signalling
2287 ** unsigned addition of a 32-bit constant with the high bit set no
2288 longer causes an internal compiler error on the x86.
2289 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2290 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2291 with negative last argument.
2292 ** byte specifiers with zero size and position no longer cause
2293 an error during type derivation.
2294 ** bignum multiplication on the Alpha platform now returns the
2296 * porting: The system now builds on SuSE AMD64, although it still
2297 generates a 32-bit binary.
2298 * .fasl file incompatibility: The fasl file version number has
2299 been incremented (because of the changes to internal compiler
2300 data structures referred to above).
2302 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2303 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2304 more accurately, on the Darwin kernel running on PowerPC hardware.
2305 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2306 * Compiler code deletion notes now signal a condition of type
2307 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2308 an associated MUFFLE-WARNING restart.
2309 * The compiler now performs limited argument count validation of
2310 constant format strings in FORMAT, and where appropriate in ERROR,
2311 CERROR and WARN. (thanks to Gerd Moellmann)
2312 * New ASDF-INSTALL contrib can be used for automatic download and
2313 installation of third-party Lisp code from CCLAN or other sites
2315 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2316 which forces another thread to execute a function supplied by the
2318 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2319 now accept and act upon their :ELEMENT-TYPE keyword argument.
2320 (reported by Martin Atzmueller, Edi Weitz)
2321 * bug fix: FILE-POSITION now accepts position designators up to
2322 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2323 is the greater. (thanks to Patrik Nordebo)
2324 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2325 Kalle Olavi Niemitalo)
2326 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2327 printable. (reported by Eric Marsden)
2328 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2329 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2330 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2331 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2332 (thanks to Stig Sandoe)
2333 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2334 UNREAD-CHAR now correctly includes the unread character in the
2335 target sequence. (thanks to Gerd Moellmann)
2336 * bug fix in threaded builds: the system can now be suspended and
2337 resumed by shell job control with minimal disruption.
2338 * bug fixes in times and timezones >2038AD
2339 * better handling of "where is GNU make?" problem in build scripts
2340 (thanks to Nikodemus Siivola)
2341 * new optimization: inside a named function any reference to a
2342 function with the same name is considered to be a self-reference;
2343 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2344 optimization quality.
2345 * new optimization on x86: logical functions and + now have
2346 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2347 used when the result is truncated to 32 bits.
2348 * VALUES declaration is partially enabled.
2349 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2350 * fixed some bugs revealed by Paul Dietz' test suite:
2351 ** The system now obeys the constraint imposed by
2352 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2353 types form a lattice under type intersection.
2354 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2355 ** ASSOC now ignores NIL elements in an alist.
2356 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2357 and (1+ MOST-POSITIVE-FIXNUM) answers.
2358 ** The addition of a method with invalid qualifiers to a generic
2359 function does not cause an error to be signalled immediately;
2360 a warning is signalled, and the error is generated only on
2361 calling the generic function.
2362 * changed .fasl file version number, in part to add type codes for
2363 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2364 obscure ANSI requirements
2366 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2367 * fixed bug 148: failure to inline-expand a local function left
2368 garbage, confusing the compiler.
2369 * fixed bugs 3cd: structure slot readers perform type check if the
2370 slot can have an invalid value (i.e. it is either not initialized
2371 or can be written with a less specific slot writer).
2372 * bug fix: the compiler now traps array references to elements off
2373 the end of an array; previously, the bounds checking in some
2374 circumstances could go off-by-one.
2375 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2377 * type declarations for array element types now obey the description
2378 on the CLHS page "Declaration TYPE", as per discussions on
2379 sbcl-help around 2003-05-08. This means that a declaration
2380 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2381 declaration, all references to BAR will be asserted or assumed
2382 (with THE, so dependent on compiler policy) to involve objects of
2383 type FOO. Note that no such declaration is implied in
2384 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2385 * declared types of functions from the "Conditions"
2386 chapter. (reported by Paul Dietz)
2387 * bug fix: CERROR accepts a function as its first argument.
2388 * bug fix: NTH an NTHCDR accept a bignum as index
2389 arguments. (reported by Adam Warner)
2390 * optimization: character compare routines now optimize comparing
2391 against a constant character. (reported by Gilbert Baumann)
2392 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2393 argument now works properly on the MIPS platform.
2394 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2395 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2396 declared to have a functional type, does not cause an error
2398 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2399 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2400 * fixed bug: if last continuation of a deleted block has a
2401 destination, this destination should be deleted too. (reported by
2403 * fixed a bug in the bootstrap process: the host compiler's values
2404 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2405 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2406 test case from Patrik Nordebo)
2407 * improved the ability of the disassembler on the PPC platform to
2408 provide helpful disassembly notes.
2409 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2410 the class in more cases than previously.
2411 * bug fix: FILE-POSITION now understands :START and :END for
2412 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2413 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2414 8) data. (thanks to David Lichteblau)
2415 * bug fix: it is possible to add a method to a generic function
2416 without lambda list.
2417 * bug fix: reader failed to signal END-OF-FILE inside an
2418 object representation. (reported by Nikodemus Siivola)
2419 * fixed some bugs revealed by Paul Dietz' test suite:
2420 ** LAST and [N]BUTLAST should accept a bignum.
2421 ** condition slot accessors are methods.
2422 ** (VECTOR NIL) is a subtype of STRING.
2424 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2425 * minor incompatible change: some nonsensical specialized lambda
2426 lists (used in DEFMETHOD) which were previously ignored now signal
2428 * minor incompatible change: the system is now aware of the types of
2429 variables in the COMMON-LISP package, and will signal errors for
2430 most violations of these type constraints (where previously they
2431 were silently accepted).
2432 * minor incompatible change: COMPILE-FILE now uses the freedom
2433 afforded (ANSI 3.2.2.3) to use derived function types for
2434 functions defined in the same file. This also permits the system
2435 to warn on static type mismatches and function
2436 redefinition. (Currently it does not work with high DEBUG level.)
2437 * minor incompatible change: VALUES declaration is disabled.
2438 * When issuing notes, the compiler now signals a condition of type
2439 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2440 restart for use in user handlers. It is expected that the
2441 COMPILER-NOTE condition will eventually become a condition
2442 supertype to a hierarchy of note types, which will then be
2443 handleable in a similar fashion. However, at the moment, no such
2444 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2445 but it's an implementation detail, not a classification for the
2447 * Changes in type checking closed the following bugs:
2448 ** type checking of unused values (192b, 194d, 203);
2449 ** template selection based on unsafe type assertions (192c, 236);
2450 ** type checking in branches (194bc).
2451 * A short form of VALUES type specifier has ANSI meaning (it has
2452 increased the number of situations when SBCL cannot perform type
2454 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2455 keywords or constants is permissible.
2456 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2457 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2458 classes. (thanks to Antonio Martinez)
2459 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2460 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2461 argument to a pretty-print stream. Instead, it leaves any such
2462 operations to DESCRIBE-OBJECT methods.
2463 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2465 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2466 lambda lists. (thanks to David Lichteblau)
2467 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2468 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2469 (thanks to Gerd Moellmann)
2470 * fixed bug 246: increased compilation speed of long
2471 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2473 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2475 * DEFCONSTANT now throws a condition of type
2476 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2477 constant to a non-EQL value; CONTINUE and ABORT restarts
2478 respectively change and preserve the value.
2479 * fixed bug 63: The code walker, part of the implementation of CLOS,
2480 is now better at handling symbol macros.
2481 * bug fix: There is no longer an internal implementation type named
2482 CL:LENGTH. (reported by Raymond Toy)
2483 * bug fix: In macro-like defining macros/special operators the
2484 implicit block does not enclose the lambda list.
2485 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2486 enclosed in parentheses) are not suitable as type specifiers, and
2487 their use properly signals an error now.
2488 * bug fix: An argument count mismatch for a type specifier in code
2489 being compiled no longer causes an unhandled error at compile
2490 time, but signals a compile-time warning.
2491 * fixed simple vector readable printing
2492 * bug fix: DESCRIBE takes more care over whether the class
2493 precedence list slot of a class is bound before accessing it.
2494 (reported by Markus Krummenacker)
2495 * bug fix: FORMATTER can successfully compile pretty-printer format
2496 strings which use variants of the ~* directive inside.
2497 * bug fix: SEARCH now applies its TEST predicate to the elements of
2498 the arguments in the correct order. (thanks to Wolfhard Buss)
2499 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2500 in type inference. (thanks to Robert E. Brown)
2501 * bug fix: Reading in symbols with an explicit package name of ""
2502 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2503 * fixed some bugs revealed by Paul Dietz' test suite:
2504 ** NIL is now allowed as a structure slot name.
2505 ** Arbitrary numbers, not just REALs, are allowed in certain
2506 circumstances in LOOP for-as-arithmetic clauses.
2507 ** Multiple class redefinitions before slot access no longer
2508 causes a type error.
2509 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2510 association between the name and a class.
2511 ** Generic functions with non-standard method-combination and over
2512 five methods all of which return constants no longer return NIL
2513 after the first few invocations. (thanks to Gerd Moellmann)
2514 ** CALL-NEXT-METHOD with no arguments now passes the original
2515 values of the arguments, even in the presence of assignment.
2516 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2518 ** DISASSEMBLE works with closures and funcallable instances.
2519 ** ADD-METHOD now returns the generic function, not the new method.
2520 ** FIND-METHOD signals an error if the lengths of the specializers
2521 is incompatible with the generic function, even if the ERRORP
2523 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2524 which its argument is a member.
2525 ** DEFCLASS only redefines the class named by its class-name
2526 argument if that name is the proper name of the class;
2527 otherwise, it creates a new class.
2528 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2529 of the UNBOUND-SLOT condition to the name of the slot.
2530 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2531 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2532 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2533 SLOT-MAKUNBOUND in the specified fashion.
2535 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2536 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2537 cross-compilation host. As a consequence, we can now bootstrap our
2538 way up to SBCL starting with a bare gcc toolchain and human-readable
2539 source code (first the source to CLISP, then the source to SBCL).
2540 * A contributed module containing a partial implementation of the
2541 simple-streams interface has been included. (thanks to Rudi
2543 * A contributed module implementing the RFC1321 Message Digest
2544 Algorithm, known as MD5, has been included.
2545 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2546 no longer has any effect, as the code controlled by this feature
2547 has been deleted. (As far as we know, no-one has ever built using
2548 this feature, and its semantics were confused in any case).
2549 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2550 work on conditions (as required by the ANSI specification),
2551 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2552 expected behaviour on conditions. Users should note, however,
2553 that such behaviour is not required by the ANSI specification,
2554 and so use of this behaviour may render their code unportable.
2555 * fixed some bugs revealed by Paul Dietz' test suite:
2556 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2558 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2559 ** SXHASH on condition objects no longer returns NIL.
2560 ** :ALLOCATION :CLASS slots are better treated; their values are
2561 updated on class redefinition, and initforms inherited from
2562 superclasses are applied.
2563 ** REMOVE-METHOD returns its generic function argument even when
2564 no method was removed.
2565 ** SHARED-INITIALIZE now initializes the values of the requested
2566 slots, including those with :ALLOCATION :CLASS.
2567 ** ALLOCATE-INSTANCE now works on structure classes defined via
2568 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2570 ** SLOT-EXISTS-P now works on conditions, as well as structures
2572 ** MAKE-LOAD-FORM now has the required methods on
2573 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2574 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2575 keyword, and now implements the SLOT-NAMES argument.
2576 ** methods with &OPTIONAL arguments no longer allow too many
2577 arguments to be passed in the call without error.
2578 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2579 option is consistent with the required arguments of the generic
2580 function lambda list.
2581 * bug fix: REQUIRE accepts a string designator. (Thanks to
2583 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2584 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2585 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2586 * bug fix: The compiler now checks for duplicated variables in macro
2588 * bug fix: SETQ on globals returns the correct value.
2589 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2590 not a COMPILER-ERROR (followed by some other strange error on
2591 choosing the CONTINUE restart).
2592 * bug fix: make.sh and friends are now more consistent in the way that
2593 they look for GNU "make".
2595 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2596 * experimental native threads support (on x86 Linux >=2.4 only).
2597 This is not compiled in by default: you need to add :SB-THREAD to
2598 the target features. See the "Beyond ANSI" chapter of the manual
2600 * fix for longstanding nonANSIism: The old distinction between
2601 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2602 The return value from CL:FIND-CLASS is now a CLOS class, and
2603 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2604 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2605 * An interface to the MetaObject Protocol, as described in Kiczales,
2606 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2607 MIT Press, 1991, is available from the SB-MOP package.
2608 * incompatible change: the SB-PCL package should now be considered
2609 a private implementation detail, and no longer a semi-private MOP
2611 * minor incompatible change: due to rearrangement for threads, the
2612 control stack and binding stack are now allocated at arbitrary
2613 addresses instead of being hardcoded per-port. Users affected by
2614 this probably have to be doing advanced things with shared
2615 libraries, and will know who they are.
2616 * minor incompatible change: Previously, all --eval forms used were
2617 processed with READ before any of them were processed with EVAL.
2618 Now each --eval form is processed with both READ and EVAL before
2619 the next --eval form is processed. (Thus package operations like
2620 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2621 work as the user might reasonably expect.)
2622 * minor incompatible change: *STANDARD-INPUT* is now only an
2623 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2625 * minor incompatible change: Y-OR-N-P is now character-oriented,
2626 not line oriented. Also, YES-OR-NO-P now works without errors.
2627 (thanks to Antonio Martinez)
2628 * sb-aclrepl module improvements: an integrated inspector, added
2629 repl features, and a bug fix to :trace command.
2630 * Known functions, which cannot be open coded by the backend, are
2631 considered to be able to check types of their arguments. (fixing
2632 a bug report by Nathan J. Froyd)
2633 * fixed a bug in computing method discriminating functions: It is
2634 now possible to define methods specialized on classes which have
2635 forward-referenced superclasses. (thanks to Gerd Moellmann)
2636 * fixed evaluation order in optional entries (reported by Gilbert
2638 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2639 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2640 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2641 required-by-AMOP NAME argument, as well as CLASS and
2642 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2643 * fixed bug 20: DEFMETHOD can define methods using names that are
2644 not the proper names of classes to designate class specializers.
2645 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2646 Unix file descriptors, instead of blowing up. (thanks to Antonio
2648 * Garbage collection refactoring: user-visible change is that a
2649 call to the GC function during WITHOUT-GCING will not do garbage
2650 collection until the end of the WITHOUT-GCING. If you were doing
2651 this you were probably losing anyway.
2652 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2653 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2654 TYPEP the latter but not the former.
2655 * The compiler issues a full WARNING for calls to undefined functions
2656 with names from the CL package.
2657 * MAP-INTO for a vector destination is open coded. (reported by
2658 Brian Downing on c.l.l)
2659 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2660 documentation string.
2661 * fixed some bugs revealed by Paul Dietz' test suite:
2662 ** COPY-ALIST now signals an error if its argument is a dotted
2664 ** Condition slots are now accessed more correctly in the presence
2665 of multiple initargs for a given slot.
2666 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2667 exclude restarts of the same name associated with a different
2669 ** DEFCLASS of forward-referenced classes with another
2670 forward-referenced class in the superclasses list no longer
2672 ** Condition slots are now initialized once each, not multiple
2673 times. (thanks to Gerd Moellmann)
2674 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2676 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2677 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2678 arguments contain duplicated elements.
2679 ** RESTART-CASE understands local macros.
2680 ** RESTART-CASE associates exactly its own restarts with a condition.
2681 ** ENDP in safe mode checks its argument to be of type LIST.
2682 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2683 in question is unbound.
2684 ** Optimized MAKE-INSTANCE functions no longer cause internal
2685 assertion failures in the presence of duplicate initargs.
2686 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2687 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2688 propagated even in the ctor optimized implementation of
2690 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2692 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2693 * a better implementation of SXHASH on (simple) bit vectors,
2694 measured both in execution speed and in distribution of results
2695 over the positive fixnums, has been installed. Likewise, a better
2696 implementation of EQUAL for simple bit vectors is now available.
2697 * fixed CEILING optimization for a divisor of form 2^k.
2698 * fixed bug 240 (emitting extra style warnings "using the lexical
2699 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2700 by Antonio Martinez)
2701 * fixed SXHASH, giving different results for NIL depending on type
2702 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2703 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2704 variable. (found by Rolf Wester)
2705 * SBCL does not ignore type declarations for special
2706 variables. (reported by rif on c.l.l 2003-03-05)
2707 * some bug fixes in contrib/sb-aclrepl/
2708 * fixed some bugs revealed by Paul Dietz' test suite:
2709 ** a bug in the CONS type specifier, whereby the CAR and CDR
2710 types got intertwined, has been fixed;
2711 ** the type system is now able to reason about the interaction
2712 between INTEGER and RATIO types more completely;
2713 ** APPEND, [N]REVERSE and NRECONC check that those their
2714 arguments, which must be proper lists, are really so;
2715 ** An array specialized to be unable to hold elements has been
2716 implemented, as required -- yes, really -- by ANSI;
2717 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2718 on malformed property lists;
2720 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2721 * incompatible packaging change: in line with Unix convention,
2722 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2723 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2724 * REQUIRE and PROVIDE are now optionally capable of doing something
2725 useful. See the documentation string for REQUIRE.
2726 * infrastructure for a managed SBCL contrib system: contributed
2727 modules in this release include:
2728 ** the ASDF system definition facility;
2729 ** an interface to the BSD Sockets API;
2730 ** an ACL-like convenience interface to the repl;
2731 (thanks to Kevin Rosenberg)
2732 ** an implementation of ROTATE-BYTE, with efficient implementation
2734 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2735 gives a closer approximation to the right answer than previously.
2736 (thanks to Raymond Toy)
2737 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2738 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2739 optional environment arguments, as required by ANSI.
2740 * fixed bugs in other functions taking environment objects, allowing
2741 calls with an explicit NIL environment argument to be compiled
2743 * fixed bug 228: primary return values from
2744 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2745 COMPILE or FUNCTION.
2746 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2747 :TYPE will no longer signal errors on innocuous objects.
2748 * fixed bug 231b: SETQ is better at respecting type declarations in
2749 the lexical environment.
2750 * fixed a bug in DEFCLASS: classes named by symbols with no or
2751 unprintable packages can now be defined.
2752 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2753 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2754 * fixed compiler failure related to checking types of functions.
2755 (reported by Robert E. Brown)
2756 * the compiler is now much more consistent in its error-checking
2757 treatment of bounding index arguments to sequence functions: in
2758 (SAFETY 3) code, errors will be signalled in almost all cases if
2759 invalid sequence bounding indices are passed to functions defined
2760 by ANSI to operate on sequences.
2761 * fixed a bug in the build procedure: documentation of SBCL-specific
2762 packages is now preserved and available in the final Lisp image.
2763 * lifted FDEFINITION lookup out of loops in the implementation of
2764 many list operations. (thanks to Robert E. Brown)
2765 * fixed a bug in the reader: the #n# reader macro now works for
2766 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2767 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2768 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2769 E. Brown and Tony Martinez respectively)
2770 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2771 SLOT-BOUNDP: in particular, it is now able to optimize them much
2772 better, and is now not vulnerable to having packages renamed.
2773 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2774 always called when a slot is not present in an instance. (thanks
2776 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2777 longer signal an error on unreading a character following EOF on
2778 the previous constituent stream. (thanks to Tony Martinez)
2779 * fixed some bugs revealed by Paul Dietz' test suite:
2780 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2781 not just nonnegative fixnums;
2782 ** the logical bit-array operators such as BIT-AND now accept an
2783 explicit NIL for their "opt-arg" argument (to indicate a
2784 freshly-consed result bit-array);
2785 ** ELT now signals an error on an invalid sequence index in safe
2787 ** the type system is now cleverer about negations of numeric
2788 types, and consequently understands the BIGNUM and RATIO types
2790 ** the type system is now cleverer about the interaction between
2791 INTEGER and RATIO types: while bugs still remain, many more
2792 cases are accurately computed;
2793 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2794 if it is in the last clause;
2795 ** CONSTANTLY now correctly returns a side-effect-free function in
2797 ** DECLARE is no longer treated as a special-operator; in
2798 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2799 * incremented fasl file version number due to the change in the
2800 DEFSTRUCT-SLOT-DESCRIPTION structure.
2802 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2803 * minor incompatible change: code processed by the "interpreter" or
2804 EVAL now has a compilation optimization policy of (DEBUG 2)
2805 (changed from (DEBUG 1)) to improve debuggability of interactive
2806 development, and to allow the use of the debug RETURN command in
2808 * an experimental implementation of the RETURN command for the
2809 debugger has been included. (thanks to Frederik Kuivinen)
2810 * fixed bug 62: constraints were not propagated into a loop.
2811 * fixed bug in embedded calls of SORT (reported and investigated by
2813 * fixed some bugs revealed by Paul Dietz' test suite:
2814 ** printing and reading of arrays with some dimensions having
2815 length 0 (thanks to Gerd Moellmann);
2816 ** BOA constructor with &AUX argument without a default value does
2817 not cause a type error;
2818 ** CONSTANTP now returns true for all self-evaluating objects.
2820 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2821 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2822 accessors that are related by inheritance, as specified in the
2823 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2824 Valtteri Vuorikoski)
2825 * The compiler is now able to inline functions that were defined in
2826 a complex lexical environment (e.g. inside a MACROLET).
2827 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2829 * Support for the upcoming FreeBSD-5.0 release has been included.
2830 (thanks to Dag-Erling Smorgrav)
2831 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2832 effect when it is not in a toplevel context.
2833 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2834 stand a better chance of being correct. (thanks to Gerd
2836 * fixed bug in COERCE, which now signals an error on coercing a
2837 rational to a bounded real type which excludes the expected
2839 * The compiler is now able to derive types more accurately from the
2840 COERCE and COMPILE functions.
2841 * fixed bug 223: functional binding is considered to be constant
2842 only for symbols in the CL package.
2843 * fixed bug 231: SETQ did not check the type of a variable being set
2844 (reported by Robert E. Brown)
2845 * A new optimization for MAKE-INSTANCE has been included, fixing
2846 various bugs (including relating to :ALLOCATION :CLASS slots and
2847 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2849 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2850 ** As required by ANSI, LOOP now disallows anonymous collection
2851 clauses such as COLLECT I in conjunction with aggregate boolean
2852 clauses such as THEREIS (= I 1);
2853 ** LOOP now signals an error when any variable is reused in the
2854 same loop (including the potentially useful construct analogous
2855 to WITH A = 1 WITH A = (1+ A);
2856 ** IT is only a special loop symbol within the first clause of a
2857 conditional loop clause;
2858 ** LOOP with a typed iteration variable over a hashtable now
2859 signals a type error iff it should.
2860 * fixed some other bugs revealed by Paul Dietz' test suite:
2861 ** FILE-STREAM now names the class previously known as FD-STREAM;
2862 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2863 argument) no longer signals an error;
2864 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2865 of the slot symbol, rather than using the current package
2866 ((:CONC-NAME "") continues to intern the slot's name in the
2868 * incremented fasl file version number, because of the incompatible
2869 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2870 of the new implementation of DEFINE-COMPILER-MACRO.
2872 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2873 * Support for building SBCL for MIPS platforms running in
2874 little-endian mode has now been checked in, and basic
2875 functionality on said platforms verified.
2876 * minor incompatible change: PCL now records the pathname of a file
2877 in which methods and the like are defined, rather than its
2879 * minor incompatible change: TRUENAME now considers the truename of
2880 a file naming a directory to be the pathname with :DIRECTORY
2881 component indicating that directory.
2882 * minor incompatible change: a NAMED clause in the extended form of
2883 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2884 reason for the previous behaviour is unclear.
2885 * more systematization and improvement of CLOS and MOP conformance
2886 in PCL (thanks to Gerd Moellman and Pierre Mai):
2887 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2889 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2890 primary methods with no specializers;
2891 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2893 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2894 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2895 FORWARD-REFERENCED-CLASSes; error reporting on
2896 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2898 ** SXHASH on CLOS instances now uses a slot internal to the
2899 instance to return different numbers on distinct instances,
2900 while preserving the same return value through invocations of
2902 ** DEFMETHOD signals errors when methods with longer incongruent
2903 lambda lists are added to generic functions;
2904 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2905 CLASS, as specified in AMOP;
2906 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2907 on the order returned by the primary method for classes of
2908 class STANDARD-CLASS;
2909 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2910 * fixed some bugs shown by Paul Dietz' test suite:
2911 ** DOLIST puts its body in TAGBODY;
2912 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2914 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2915 value producing form;
2916 ** if more variables are given to PROGV than values, extra
2917 variables are bound and made to have no value;
2918 ** NSUBSTITUTE on list arguments gets the right answer with
2920 ** ELT signals an error of type TYPE-ERROR when the index argument
2921 is not a valid sequence index;
2922 ** LOOP signals (at macroexpansion time) an error of type
2923 PROGRAM-ERROR when duplicate variable names are found;
2924 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2925 ** FUNCALL of special-operators now cause an error of type
2927 ** PSETQ now works as required in the presence of side-effecting
2928 symbol-macro places;
2929 ** NCONC accepts any object as its last argument;
2930 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2932 ** loop-for-as-package does not require a package to be explicitely
2934 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2936 * fixed bug 166: compiler preserves "there is a way to go"
2937 invariant when deleting code.
2938 * fixed bug 172: macro lambda lists with required arguments after
2939 &REST arguments now cause an error to be signalled. (thanks to
2941 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2942 bug. (thanks to Matthew Danish)
2943 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2945 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2946 arguments are lexically rebound. (thanks to Gerd Moellmann and
2948 * fixed bug 194: error messages are now more informative when there
2949 is no primary method applicable in a call to a generic
2950 function. (thanks to Gerd Moellmann)
2951 * fixed bug in command line argument checking (thanks to Julian
2953 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2954 (thanks to Matthew Danish)
2955 * incremented fasl file version number, because of the
2956 SXHASH-related changes in the layout of CLOS data structures
2958 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2959 * minor incompatible change: The runtime (the Unix executable named
2960 "sbcl") is now much pickier about the .core files it will load.
2961 Essentially it now requires .core files to descend from the same
2962 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2963 as the runtime does. (The intent is to prevent the crashes which
2964 can occur, and which can even be reported as mysterious failures,
2965 when people patch the sources or change the build parameters
2966 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2967 sbcl and .core files.)
2968 * fixed bug: VALUES-LIST is no longer optimized away.
2969 * fixed bug 142: The FFI conversion of C string values to Lisp
2970 string values no longer conses excessively. (thanks to Nathan
2971 Froyd porting Raymond Toy's fix to CMU CL)
2972 * began to systematize and improve MOP conformance in PCL (thanks to
2973 Nathan Froyd, Gerd Moellman and Pierre Mai):
2974 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2976 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2977 ** FINALIZE-INHERITANCE is now called on class finalization;
2978 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2979 argument precedence order.
2980 * fixed bug 202: The compiler no longer fails on functions whose
2981 derived types contradict their declared type.
2982 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2983 so it can be non-toplevel.
2984 * The fasl file version number has changed (because of the new
2985 implementation of DEFMACRO).
2986 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2987 safe code, that any length requirement by their type-specifier
2988 argument is valid. The exceptions to this are described in bug
2990 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2991 functions now check the types of their inputs as required by ANSI.
2992 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2993 introduced symbol is DECLAREd to be SPECIAL.
2994 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2995 * fixed bug: :COUNT argument to sequence functions may be negative
2996 * fixed bug: body of DO-SYMBOLS may contain declarations
2997 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2998 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2999 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3000 symbol macro only once
3001 * fixed printing of call frame when argument list is unavailable
3002 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3003 * compiler no longer signals WARNING on unknown keyword
3006 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3007 * A beta-quality port to the mips architecture running Linux,
3008 based on the old CMUCL backend, has been made. It has been tested
3009 on a big-endian kernel, and works sufficiently well to be able to
3010 rebuild itself; it has not been tested in little-endian mode.
3011 * fixed an inconsistency between gencgc.c and purify.c which made
3012 dumping/loading .core files unreliable
3013 * fixed bug 120a: The compiler now deals correctly with IFs where
3014 the consequent is the same as the alternative, instead of
3015 misderiving the return type. (thanks to Alexey Dejneka)
3016 * fixed bug 113: Logical pathnames are now dumpable (the logical
3017 host is resolved at load-time, throwing an error if it is not
3019 * fixed bug 174: FORMAT's error message is slightly clearer when a
3020 non-printing character is used in a format directive.
3021 * fixed several bugs in compiler checking of type declarations, i.e.
3022 violations of the Python "declarations are assertions" principle
3023 (thanks to Alexey Dejneka)
3024 * fixed several bugs in PCL's error checking (thanks to Gerd
3026 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3028 * fixed bug in compilation of functions as first class values
3029 (thanks to Antonio Martinez-Shotton)
3030 * The compiler's handling TYPE-ERRORs which it can prove will
3031 inevitably happen at runtime has been cleaned up and corrected
3032 in several ways. (thanks to Alexey Dejneka)
3033 * improved argument type checking for various basic arithmetic
3034 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3035 lavished on them in the past that they can be compiled in many
3036 ways in different special cases
3037 * fixed bug 181: compiler checks validity of user supplied type
3039 * cleaned up code flushing in optimization: Function calls which
3040 should signal errors for safety purposes (e.g. which ANSI says
3041 should signal errors when their arguments are of incorrect type)
3042 are no longer optimized away.
3043 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3044 * incremented fasl file version number, because changes in the
3045 implementation of sequence functions like COERCE caused
3046 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3047 in old inline expansions) to become undefined. (Actually these
3048 changes were later undone, so we might very well be binary
3049 compatible with 0.7.7 after all, but leaving the version number
3050 incremented seemed like the simplest and most conservative
3053 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3054 * An alpha-quality port to the parisc architecture running Linux,
3055 based on the old CMUCL backend, has been made. This, even more so
3056 than the other backends, should be considered still a work in
3057 progress; known problems include that the Linux kernel in 64-bit
3058 mode does not propagate the correct sigcontext structure to
3059 userspace, and consequently SBCL on a parisc64 kernel will not
3061 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3062 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3063 suppressed.) Also now that the compiler is looking at declarations
3064 in the environment, it checks optimization declarations as well,
3065 and suppresses inlining when (> DEBUG SPEED).
3066 * More fixes have been made to treatment of floating point exception
3067 treatment and other Unix signals. In particular, floating point
3068 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3069 * The detection and handling of control stack exhaustion (infinite
3070 or very deeply nested recursion) has changed. Stack exhaustion
3071 detection is now done by write-protecting pages at the OS level
3072 and applies at all optimization settings; when found, a
3073 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3074 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3075 be caught using IGNORE-ERRORS.
3076 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3077 correctly when transforming recursive LABELS functions to LETs.
3078 (thanks to Alexey Dejneka)
3079 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3080 that are names of constants or global variables.
3081 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3082 alien routines with docstrings.
3083 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3084 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3086 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3087 (thanks to Eric Marsden)
3088 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3089 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3090 * Bug fix: The SPARC backend can now compile functions involving
3091 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3092 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3093 to a functions expecting a PATHNAME-DESIGNATOR.
3094 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3095 lambda lists. (thanks to Alexey Dejneka)
3096 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3097 functions so that EXPORTing the name of the class doesn't cause
3098 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3099 to Antonio Martinez for reporting this)
3100 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3101 constant-folded without failing an assertion. (thanks to Einar
3102 Floystad Dorum for reporting this)
3103 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3104 types are now dealt with more correctly by the compiler.
3105 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3106 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3107 behaviour analogous to RENAME-FILE. This puts its behaviour more
3108 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3110 * The fasl file version number has changed again. (because of the
3111 bug fix involving the names of PCL MAKE-INSTANCE functions)
3113 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3114 * bug fix: Floating point exceptions are treated much more
3115 consistently on the x86/Linux and PPC/Linux platforms.
3116 * Array initialization with :INITIAL-ELEMENT is now much faster for
3117 cases when the compiler cannot open code the array creation, but
3118 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3119 array accesses have also seen a speed increase.
3120 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3121 specified. (This was at the root of some bad interactions between
3122 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3124 * bug fix: Internal error arguments for undefined functions are now
3125 computed correctly on the PPC/Linux platform.
3126 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3127 Raymond Toy's patch for CMU CL)
3128 * Support for the Solaris 9 operating environment has been included
3129 (thanks to Daniel Merritt)
3130 * A very ugly but hopefully complete draft of the missing FFI chapter
3131 of the manual has been created by reformatting the corresponding
3132 CMU CL manual chapter into (currently very ugly and incoherent)
3133 DocBook and bringing it up to date for SBCL behavior. Thus, the
3134 manual is now essentially complete, at least by my extreme
3135 once-and-only-once standards, whereby it's acceptable to refer to
3136 the doc strings of SB-EXT functions as the primary documentation.
3137 * The fasl file version number has changed again, due to cleanup of
3138 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3139 is no longer a static symbol.)
3141 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3142 * SBCL now builds with OpenMCL (version 0.12) as the
3143 cross-compilation host; also, more progress has been made toward
3144 bootstrapping under CLISP.
3145 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3147 * bug 158 fixed: The compiler can now deal with integer loop
3148 increments different from 1; fixing this turned out also to fix
3150 * bug 169 fixed: no more bogus warnings about using lexical bindings
3151 despite the presence of perfectly good SPECIAL declarations (thanks
3152 to David Lichteblau)
3153 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3154 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3155 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3157 * bug 180 fixed: Method combination specifications no longer ignore
3158 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3159 * bug fix: Structure type predicate functions now check their argument
3160 count as they should.
3161 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3162 correctly. (thanks to Pierre Mai)
3163 * minor incompatible change: The --noprogrammer option is deprecated
3164 in favor of the new --disable-debugger option, which is very similar.
3165 (The major difference is that it takes effect at a slightly different
3166 time at startup, causing handling of errors in --sysinit and
3167 --userinit files will be affected differently.) The
3168 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3169 been added to allow this functionality to be controlled from ordinary
3170 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3171 maintainers, who might want to run non-interactive scripts to
3172 build SBCL cores which will later be used interactively.)
3173 * minor incompatible change: The LOAD function no longer, when given
3174 a wild pathname to load, loads all files matching that pathname.
3175 Instead, an error of type FILE-ERROR is signalled.
3177 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3178 * bug 147 fixed: The compiler preserves its block link/count
3179 invariants more correctly now so that it doesn't crash. (thanks
3181 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3183 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3184 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3185 no-value-for-this-character as NIL (as specified by ANSI).
3186 * HOST-NAMESTRING on physical pathnames now returns a string that is
3187 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3188 (thanks to Christophe Rhodes)
3189 * The Alpha port handles icache flushing more correctly. (thanks to
3191 * More progress has been made toward bootstrapping under CLISP. (thanks
3192 to Christophe Rhodes)
3193 * The fasl file format has changed again, because dynamic loading
3194 on OpenBSD (which has non-ELF object files) motivated some cleanups
3195 in the way that foreign symbols are transformed and passed around.
3196 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3197 is no longer treated as whitespace by the reader, but instead as
3198 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3199 |A
\7fB|, instead of A as it used to.
3201 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3202 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3203 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3204 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3205 this as of 0.7.1.45, but was left out of the previous news section
3206 (thanks to Dan Barlow)
3207 * SBCL now runs on the Solaris operating system on SPARC architectures
3208 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3209 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3210 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3211 * SPARC backend cleanups, allowing builds of cores optimized for V8
3212 and V9 SPARCS, and also emission of code targeted to a particular
3213 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3215 * SBCL is closer to bootstrapping under CLISP, thanks to various
3216 fixes by Christophe Rhodes.
3217 * The fasl file format has changed again, to allow the compiler's
3218 INFO database to support symbol macros.
3219 * The user manual (in doc/) is formatted into HTML more nicely.
3220 (thanks to coreythomas)
3221 * The system is smarter about SUBTYPEP relationships, especially
3222 those involving NOT types (including types such as ATOM which are
3223 represented internally using NOT types). Thus SUBTYPEP is less
3224 likely to return (VALUES NIL NIL) in general, and in particular
3225 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3227 * The fasl file format has changed again, because the internal
3228 representation of types now includes a new slot to support the new
3229 SUBTYPEP-of-NOT-types logic.
3230 * (not a change in the main branch of SBCL, but a related prototype
3231 which can hopefully be merged into the main branch of SBCL in the
3232 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3233 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3234 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3235 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3236 dispatch macro character combinations. (thanks to Alexey Dejneka)
3237 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3238 that superficially look like logical namestrings correctly.
3239 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3241 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3242 default supplied-p) for &optional and &key arguments. (thanks to
3245 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3246 * incompatible change: The compiler is now less aggressive about
3247 tail call optimization, doing it only when (> SPACE DEBUG) or
3248 (> SPEED DEBUG). (This is an incompatible change because there are
3249 programs which relied on the old CMU-CL-style behavior to optimize
3250 away their unbounded recursion which will now die of stack overflow.)
3251 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3252 for non-GENCGC systems has been increased to 20M (since that
3253 seems much closer to the likely performance optimum for modern
3254 systems than the old 4M value was)
3255 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3256 and *DEBUG-PRINT-LEVEL*
3257 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3258 of CMU CL's support for SPARC, and various endianness and other
3259 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3260 * new syntactic sugar for the Unix command line: --load foo.bar is now
3261 an alternate notation for --eval '(load "foo.bar")'.
3263 ** The system now detects stack overflow and handles it gracefully,
3264 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3265 optimization settings. (This is a good thing in general, and
3266 its introduction in this version should be particularly timely
3267 for anyone whose code fails because of suppression of tail
3269 ** The system now hunts for the C variable "environ" in a more
3270 devious way, to avoid segfaults when the C library version differs
3271 between compile time and run time. (thanks to Christophe Rhodes)
3272 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3273 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3274 ** The compiler no longer issues bogus style warnings for undefined
3275 classes in the same source file as the DEFCLASSes which defined
3276 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3278 ** fixes in CONDITION class precedence list for undefined function
3279 errors (thanks to Alexei Dejneka)
3280 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3281 correctly. (thanks to Dan Barlow)
3282 ** portability fixes aiming at bootstrapping under CLISP (thanks
3283 to Dave McDonald and Christophe Rhodes)
3284 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3285 ** fixes in type translation and and type inference (thanks to
3287 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3288 ** various fixes in the new ports (thanks to Dan Barlow)
3289 * several changes related to debugging:
3290 ** suppression of tail recursion, as noted above
3291 ** stack overflow detection, as noted above
3292 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3293 is now the default. (For some time encapsulation has been more
3294 reliable than the breakpoint-based :ENCAPSULATE NIL
3295 implementation, at least on X86 systems; and I just noticed that
3296 encapsulation also seems closer to the spirit of the ANSI
3299 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3301 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3302 up properly again. (There was a packaging bug in 0.7.0 which
3303 left their definitions in SB-SYS::LOAD-FOREIGN and
3304 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3305 vital for most things which interface to C-level interfaces,
3306 like extensions working with sockets or databases or
3307 Perl-compatible regexes or whatever, and the need to fix
3308 this bug is the main reason that 0.7.1 was released so
3310 ** DEFGENERIC is now choosier about the methods it redefines, so that
3311 reLOADing a previously-LOADed file containing DEFGENERICs does
3312 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3313 cycle now works as it should. (thanks to Alexey Dejneka)
3314 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3315 fixed by Christophe Rhodes. (He actually submitted this patch
3316 months ago, and I delayed until after 0.7.0.)
3317 ** Bug 111 (internal compiler confusion about runtime checks on
3318 FUNCTION types) was fixed by Alexey Dejneka.
3319 * Some internal cleanups (getting rid of variables which aren't
3320 needed now that the byte interpreter is gone) caused the fasl
3321 file format number to change again.
3323 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3324 * major incompatible change: The default fasl file extension, i.e. the
3325 default extension for files produced by COMPILE-FILE, has changed
3326 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3328 ** There are many changes in the implementation of the compiler.
3329 SBCL is now essentially a compiler-only implementation of ANSI
3330 Common Lisp. EVAL still "interprets" a few special cases, but
3331 almost all the interesting cases are handled by creating
3332 a LAMBDA expression, calling COMPILE on it, then calling
3333 FUNCALL on the result.
3334 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3335 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3336 Since the code is newer, there might still be some new bugs
3337 (though not as many as before Martin Atzmueller's fixes:-). But
3338 the new code is substantially simpler and clearer, and hopefully
3339 any remaining bugs will be simpler, less fundamental, and more
3340 fixable then the bugs in the old code.
3341 ** The revised compiler is still a little unsteady on its feet.
3343 *** The debugging information it produces (particularly the names
3344 of FUNCTION objects) is sometimes much less useful than what
3345 the old compiler produced.
3346 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3347 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3348 within a MACROLET) has been temporarily weakened.
3349 ** There are new compiler optimizations for various functions:
3350 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3351 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3352 *** the math functions TRUNCATE, FLOOR, and CEILING
3353 *** the function-of-all-trades COERCE
3354 Mostly these should be transparent, but there's one
3355 potentially-annoying problem (bug 117): when the compiler
3356 inline-expands a function and does type analysis on the result,
3357 it can create control paths which have type mismatches, and
3358 when it can't prove that those control paths aren't taken,
3359 it will issue WARNINGs about the type mismatches. This is
3360 a particular problem in practice for the new sequence functions.
3361 It's not clear how this should be fixed, and for now, a
3362 workaround is given in the entry for 117 in the BUGS file.
3363 ** (Because of the interaction between the two previous items --
3364 occasional inlining problems and new inline expansions -- some
3365 of the new sequence function optimizations won't really kick in
3366 completely until debugging information, and then inlining, are
3367 straightened out in some future version.)
3368 * minor incompatible changes:
3369 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3370 DIRECTORY no longer implicitly promotes NIL slots of its
3371 pathname argument to :WILD. In particular, when you ask for the
3372 contents of a directory (which you used to be able to do without
3373 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3374 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3375 ** changes in behavior that ANSI explicitly defines to be
3376 implementation dependent:
3377 *** The new compiler-only implementation still conforms with ANSI,
3378 but acts a little different than before. Besides the obvious
3379 changes in performance tradeoffs (that the cost per form passed
3380 to EVAL has gone up, and the cost per form executed by EVAL
3381 has gone down), the behavior of the system changes a little
3382 because there are no longer any interpreted function objects.
3383 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3384 e.g. doing COMPILE on the output of interactive DEFUN is
3386 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3388 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3389 has changed from (:CL) to NIL.
3390 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3391 e.g. #\Newline and #\Tab, don't have names specified in the
3392 ANSI Common Lisp standard, is now based on their ASCII symbolic
3393 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3394 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3395 are no longer used for output.
3396 ** changes in internal implementation constants:
3397 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3398 4 million. (If your application spends a lot of time GCing and
3399 you have a lot of RAM, you might want to experiment with
3400 increasing it even more.)
3401 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3402 However, almost all old code should still continue to work without
3403 immediate update, as SB-C-CALL is now a (deprecated) nickname
3405 ** Old operator names in the style DEF-FOO are now deprecated in
3406 favor of new corresponding names DEFINE-FOO, for consistency with
3407 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3408 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3409 internal symbols, but a few supported extensions like
3410 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3411 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3412 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3413 etc. as you get deeper into recursive calls to the debugger
3414 command loop, instead of the old "5]", "5]]", "5]]]"
3415 sequence. (I was motivated to do this when squabbles between
3416 ILISP and SBCL left me very deeply nested in the debugger. In the
3417 short term, this change will probably provoke more ILISP/SBCL
3418 squabbles, but hopefully it will be an improvement in the long run.)
3419 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3420 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3421 function, since declaiming return types involving aliens is
3422 (1) annoyingly messy to do by hand and (2) vital to efficient
3423 compilation of code which calls such functions.
3424 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3425 longer reexported by the SB-EXT package. They're solely useful
3426 for alien code, so it seems more logical that you should get
3427 them from the SB-ALIEN package, not in SB-EXT.
3428 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3429 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3430 features. Instead, the code that they used to control is always
3431 built into the system.
3432 * many other bug fixes
3433 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3434 advantage of the new EVAL-WHEN stuff and to clean them up in
3435 general, and they are now more ANSI-compliant in a number of
3436 ways. Martin Atzmueller is responsible for a lot of this.
3437 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3439 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3440 *** correct ERROR type for various file operations
3441 *** some fixes for Lisp streams
3442 *** DEFMETHOD syntax checking
3443 *** changing old weird representation of debug information as
3444 strings (which, among their other deficiencies, don't transform
3445 correctly when you rename packages, and don't change their
3446 print representation when you change things like *PACKAGE*
3447 and *PRINT-LENGTH*) to symbols and lists of symbols
3448 He also made several improvements and fixed several bugs in DESCRIBE.
3449 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3451 *** misbehavior of WRITE-STRING/WRITE-LINE
3452 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3453 and several other LOOP problems as well
3454 *** DIRECTORY when similar filenames are present
3455 *** DEFGENERIC with :METHOD options
3456 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3457 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3458 *** argument ordering in FIND with :TEST option
3459 *** mishandled package designator argument in APROPOS-LIST
3460 *** various problems in the backquote readmacro
3461 *** a bug in APROPOS
3462 *** probably some others that I'm not describing very well here,
3463 since the CVS log documents them by reference to sbcl-devel
3464 messages, and the SourceForge archives aren't working well.:-(
3465 ** Dan Barlow improved the Alpha port (and is making progress on the
3466 PPC port, for those of you who think different).
3467 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3468 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3469 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3470 continues to work on the SPARC port (for those of you in a position
3471 to look down upon our little PC-compatible boxes from a great height).
3472 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3473 and uses that copy, rather than the current dynamic value, when
3474 it's trying to decide whether to truncate output. Thus e.g.
3475 (let ((*print-lines* 50))
3476 (pprint-logical-block (stream nil)
3478 (let ((*print-lines* 8))
3479 (print (aref possiblybigthings i) stream)))))
3480 should now truncate the logical block only at 50 lines, instead of
3481 often truncating it at 8 lines, as it did before.
3482 * The doc/cmucl/ directory, containing old CMU CL documentation
3483 from the time of the fork, is no longer part of the base system.
3484 SourceForge has shut down its anonymous FTP service, and with it
3485 my original plan for distributing the old CMU CL documentation
3486 there. For now, if you need these files you can download an old
3487 SBCL source release and extract them from it.
3488 * The fasl file version number changed again, for dozens of reasons,
3489 some of which are apparent above.
3491 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3492 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3493 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3494 patch, so that DISASSEMBLE output is much nicer.
3495 * The code in the SB-PROFILE package now seems reasonably stable.
3496 I still haven't decided what the final interface should look like
3497 (I'd like PROFILE to interact cleanly with TRACE, since both
3498 facilities use function encapsulation) but if you have a need
3499 for profiling now, you can probably use it successfully with
3500 the current CMU-CL-style interface.
3501 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3502 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3503 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3504 link now returns the dangling link itself, and for similar
3505 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3506 link itself. (In these cases the old code signalled an error and
3507 looped endlessly, respectively.) Thus, DIRECTORY now works even
3508 in the presence of dangling and cyclic symbolic links.
3509 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3510 is now a supported extension again, since the consensus on
3511 sbcl-devel was that it can be useful for ordinary development
3512 work, not just for debugging SBCL itself.
3513 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3514 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3515 that currently-defined functions might be redefined later with
3516 different return types.
3517 * Hash tables can be printed readably, as inspired by CMU CL code
3518 of Eric Marsden and SBCL code of Martin Atzmueller.
3519 * better error handling in CLOS method combination, thanks to
3520 Martin Atzmueller porting Pierre Mai's CMU CL patches
3521 * more overflow fixes for >16Mbyte I/O buffers
3522 * A bug in READ has been fixed, so that now a single Ctrl-D
3523 character suffices to cause end-of-file on character streams.
3524 In particular, now you only need one Ctrl-D at the command
3525 line (not two) to exit SBCL.
3526 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3528 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3529 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3530 does the right thing.
3531 * The implementation of some type tests, especially for CONDITION
3532 types, is now tidier and maybe faster, due to CMU CL code
3533 originally by Douglas Crosher, ported by Martin Atzmueller.
3534 * Some math functions have been fixed, and there are new
3535 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3536 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3537 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3538 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3539 likely to be built into sbcl-0.7.0.)
3540 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3541 CL, ported to SBCL by Martin Atzmueller)
3542 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3543 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3544 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3545 * There's a new slam.sh hack to shorten the edit/compile/debug
3546 cycle for low-level changes to SBCL itself, and a new
3547 :SB-AFTER-XC-CORE target feature to control the generation of
3548 the after-xc.core file needed by slam.sh.
3549 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3550 COMPILE-FILE is no longer supported, so that now every function
3551 gets an entry point, so that block compilation looks a little
3552 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3553 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3554 returns the number of bytes consed since the system started,
3555 rather than the number consed since the first time the function
3556 was called. (The new definition parallels ANSI functions like
3557 CL:GET-INTERNAL-RUN-TIME.)
3558 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3559 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3560 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3561 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3562 * Fasl file version numbers are now independent of the target CPU,
3563 since historically most system changes which required version
3564 number changes have affected all CPUs equally. Similarly,
3565 the byte fasl file version is now equal to the ordinary
3568 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3569 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3570 is no longer recognized. I apologize for this, because it was
3571 listed in SB-EXT as a supported extension, but I found that
3572 its existing behavior was poorly specified, as well as incorrectly
3573 specified, and it looked like too much of a mess to straighten it
3574 out. I have enough on my hands trying to get ANSI stuff to work..
3575 * many patches ported from CMU CL by Martin Atzmueller, with
3576 half a dozen bug fixes in pretty-printing and the debugger, and
3577 half a dozen others elsewhere
3578 * fixed bug 13: Floating point infinities are now supported again.
3579 They might still be a little bit flaky, but thanks to bug reports
3580 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3581 as flaky as they were.
3582 * The --noprogrammer command line option is now supported. (Its
3583 behavior is slightly different in detail from what the old man
3584 page claimed it would do, but it's still appropriate under the
3585 same circumstances that the man page talks about.)
3586 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3587 are now supported, and enabled by default. Thus, the compiler can
3588 handle many floating point and complex operations much less
3589 inefficiently. (Thus e.g. you can implement a complex FFT
3591 * The compiler now detects type mismatches between DECLAIM FTYPE
3592 and DEFUN better, and implements CHECK-TYPE more correctly, and
3593 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3594 thanks to patches from Martin Atzmueller.
3595 * various fixes to make the cross-compiler more portable to
3596 ANSI-conforming-but-different cross-compilation hosts (notably
3597 Lispworks for Windows, following bug reports from Arthur Lemmens)
3598 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3599 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3600 fixed thanks to Pierre Mai's CMU CL patches.
3601 * fixes to keep the system from overflowing internal counters when
3602 it tries to use i/o buffers larger than 16M bytes
3603 * fixed bug 45a: Various internal functions required to support
3604 complex special functions have been merged from CMU CL sources.
3605 (When I was first setting up SBCL, I misunderstood a compile-time
3606 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3607 * improved support for type intersection and union, fixing bug 12
3608 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3609 more obscure bugs as well
3610 * some steps toward byte-compiling non-performance-critical
3611 parts of the system, courtesy of patches from Martin Atzmueller
3612 * Christophe Rhodes has made some debian packages of sbcl at
3613 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3614 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3615 stable, but are nonetheless usable. When he's ready, I'd be happy
3616 to add them to the SourceForge "File Releases" section. (And if
3617 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3618 * new fasl file format version number (because of changes in
3619 internal representation of (OR ..) types to accommodate the new
3620 support for (AND ..) types, among other things)
3622 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3623 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3625 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3627 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3628 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3629 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3630 * DESCRIBE now works on structure objects again.
3631 * Most function call argument type mismatches are now handled as
3632 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3633 know whether the function will be redefined before the call is
3634 executed. (The compiler could flag local calls with full WARNINGs,
3635 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3636 it doesn't keep track of enough information to know whether calls
3637 are local in this sense.)
3638 * Compiler output is now more verbose, with messages truncated
3639 later than before. (There should be some supported way for users
3640 to override the default verbosity, but I haven't decided how to
3641 provide it yet, so this behavior is still controlled by the internal
3642 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3643 src/compiler/ir1util.lisp.)
3644 * Fasl file format version numbers have increased again, because
3645 support for the Gray streams extension changes the layout of the
3646 system's STREAM objects.
3647 * The Gray subclassable streams extension now works, thanks to a
3648 patch from Martin Atzmueller.
3649 * The full LOAD-FOREIGN extension (not just the primitive
3650 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3651 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3652 but like most other programs, it defaults to copying the Unix
3653 environment from the original process instead of starting the
3654 new process in an empty environment.
3655 * Extensions which manipulate the Unix environment now support
3656 an :ENVIRONMENT keyword option which doesn't smash case or
3657 do other bad things. The CMU-CL-style :ENV option is retained
3658 for porting convenience.
3659 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3660 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3662 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3664 * A patch from Martin Atzmueller seems to have solved the SIGINT
3665 problem, and as far as we know, signal-handling now works cleanly.
3666 (If you find any new bugs, please report them!)
3667 * The system no longer defaults Lisp source file names to types
3668 ".l", ".cl", or ".lsp", but only to ".lisp".
3669 * The compiler no longer uses special default file extensions for
3670 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3671 seems to expect a single default extension for all compiled code,
3672 and there's no compelling reason to try to stretch the standard
3673 to allow two different extensions.) Instead, byte-compiled files
3674 default to the same extension as native-compiled files.
3675 * Fasl file format version numbers have increased again, because
3676 a rearrangement of internal implementation packages made some
3677 dumped symbols in old fasl files unreadable in new cores.
3678 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3679 many fewer weird special cases.
3680 * Bug #17 (differing COMPILE-FILE behavior between logical and
3681 physical pathnames) has been fixed, and some related misbehavior too,
3682 thanks to a patch from Martin Atzmueller.
3683 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3684 by Tim Moore, ported to SBCL by Martin Atzmueller.
3685 * Martin Atzmueller fixed several filesystem-related problems,
3686 including bug #36, in part by porting CMU CL patches, which were
3687 written in part by Paul Werkowski.
3688 * More compiler warnings in src/runtime/ are gone, thanks to
3689 more patches from Martin Atzmueller.
3690 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3693 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3695 * DESCRIBE now works on CONDITION objects.
3696 * The debugger now handles errors which arise when trying to print
3697 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3699 * The build system now uses an additional file, customize-target-features.lisp,
3700 to allow local modifications to the target *FEATURES* list. (The point of
3701 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3702 debugging features enabled, without having to worry about propagating it
3703 into everyone's system when I do a "cvs update".) When no
3704 customize-target-features.lisp file exists, the target *FEATURES* list
3705 should be constructed the same way as before.
3706 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3707 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3708 warn about behavior which is undefined under the ANSI spec. Note: This
3709 is specified by ANSI, but it's not very popular with programmers.
3710 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3711 macro in the SBCL sources for an example of a workaround which you
3712 might use to make portable ANSI-standard code which does what you want.
3713 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3714 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3715 when it's used as a non-toplevel form. (This is required in order
3716 to implement the DEFCONSTANT-EQX macro.)
3717 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3718 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3719 funny things, due to bugs in the implementation of EVAL-WHEN
3720 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3721 non-toplevel DEFCONSTANTs are uncommon.)
3722 * The core file version number and fasl file version number have been
3723 incremented, because the old noncompliant DEFCONSTANT behavior involved
3724 calling functions which no longer exist, and because I also took the
3725 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3726 * fixed bug 1 (error handling before read-eval-print loop starts), and
3727 redid debugger restarts and related debugger commands somewhat while
3729 ** The QUIT debugger command is gone, since it did something
3730 rather different than the SB-EXT:QUIT command, and since it never
3731 worked properly outside the main toplevel read/eval/print loop.
3732 Invoking the new TOPLEVEL restart provides the same functionality.
3733 ** The GO debugger command is also gone, since you can just invoke
3734 the CONTINUE restart directly instead.
3735 ** The TOP debugger command is also gone, since it's redundant with the
3736 FRAME 0 command, and since it interfered with abbreviations for the
3738 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3740 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3741 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3742 some porting work by Martin Atzmueller)
3743 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3744 with a boatload of patches which he ported to SBCL. Now that those
3746 ** The system tries to make sure that its low-priority messages
3747 are prefixed by semicolons, to help people who like to use
3748 syntax highlighting in their ILISP buffer. (This patch
3749 was originally due to Raymond Toy.)
3750 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3751 patches originally written by Raymond Toy.
3752 ** The compiler understands coercion between single-value and
3753 multiple-VALUES type expressions better, getting rid of some very
3754 weird behavior, thanks to patches originally by Robert MacLachlan
3755 and Douglas Crosher.
3756 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3757 lambda lists, thanks to a patch originally by Pierre Mai.
3758 ** The system no longer bogusly warns about "abbreviated type
3760 ** The compiler gets less confused by inlining and RETURN-FROM,
3761 thanks to some patches originally by Tim Moore.
3762 ** The system no longer hangs when dumping circular lists to fasl
3763 files, thanks to a patch originally from Douglas Crosher.
3764 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3765 undefined function error.
3766 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3767 in structure slot initforms) for now, documented workaround instead:-|
3768 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3769 * fixed bug 5: added stubs for various Gray stream functions called
3770 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3771 installed, at least appropriate type errors are generated
3772 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3773 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3774 consistently in DEFMETHOD forms.
3775 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3776 it doesn't seem to affect SBCL after all
3777 * The C runtime system now builds with better optimization and many
3778 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3780 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3782 * The system is now under CVS at SourceForge (instead of the
3783 CVS repository on my home machine).
3784 * The new signal handling code has been tweaked to treat register
3785 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3786 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3787 downstream, e.g. in debug-int.lisp, has implicit dependencies
3788 on the unsignedness of integer representation of machine words,
3789 and that caused the system to bomb out with infinite regress
3790 when trying to recover from type errors involving signed values,
3791 e.g. (BUTLAST '(1 2 3) -1).)
3792 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3793 (This was one of the bugs Peter Van Eynde reported back in July.)
3794 * The system now uses code inspired by Colin Walters' O(N)
3795 implementation of MAP (from the cmucl-imp@cons.org mailing
3796 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3797 inline the MAP operation, and there is more than one
3798 sequence argument to the MAP call (so that it can't just
3799 do ETYPECASE once and for all based on the type of the
3800 single sequence argument). (The old non-inline implementation
3801 of the general M-argument sequence-of-length-N case required
3802 O(M*N*N) time when any of the sequence arguments were LISTs.)
3803 * The QUIT :UNIX-CODE keyword argument has been renamed to
3804 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3805 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3807 * Raymond Wiker's patches to port dynamic loading from Linux to
3808 FreeBSD have been added.
3809 * The BUGS file is now more nearly up to date, thanks in large part
3810 to Martin Atzmueller's review of it.
3811 * The debugger now flushes standard output streams before it begins
3812 its output ("debugger invoked" and so forth).
3813 * The core version number and fasl file version number have both
3814 been incremented, because of incompatible changes in the layout
3816 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3817 used to not be called for a saved Lisp image.)
3818 * Martin Atzmueller's version of a patch to fix a compiler crash,
3819 as posted on sbcl-devel 13 September 2000, has been installed.
3820 * Instead of installing Martin Atzmueller's patch for the
3821 compiler transform for SUBSEQ, I deleted the compiler transform,
3822 and transforms for some similar consing operations.
3823 * A bug in signal handling which kept TRACE from working on OpenBSD
3825 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3826 away by constant folding
3827 * The system now defines its address space constants in one place
3828 (in the Lisp sources), and propagates them automatically elsewhere
3829 (through GENESIS and the sbcl.h file). Therefore, patching the
3830 address map is less unnecessarily tedious and error-prone. The
3831 Lisp names of address space constants have also been systematized.
3832 * CVS tags like dollar-Header-dollar have been removed from
3833 the sources, because they have never saved me trouble and
3834 they've been source of trouble working with patches and other
3835 diff-related operations.
3836 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3837 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3839 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3841 * The system has been ported to OpenBSD.
3842 * The system now compiles with a simple "sh make.sh" on the systems
3843 that it's supported on. I.e., now you no longer need to tweak
3844 text in the target-features.lisp-expr and symlinks in src/runtime/
3845 by hand, the make.sh takes care of it for you.
3846 * The system is no longer so grossly inefficient when compiling code
3847 involving vectors implemented as general (not simple) vectors (VECTOR T),
3848 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3849 which dares to use the various sequence functions on non-simple
3850 vectors, takes less of a performance hit.
3851 * There is now a primitive type predicate VECTOR-T-P
3852 to test for the (VECTOR T) type, so that e.g.
3853 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3854 can now be compiled with some semblance of efficiency. (The old code
3855 turned the type declaration into a full call to %TYPEP at runtime!)
3856 * AREF on (VECTOR T) is still not fast, since it's still compiled
3857 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3858 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3859 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3860 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3861 without making full calls to SUBTYPEP at runtime.
3862 (Some analogous efficiency issues for non-simple vectors specialized to
3863 element types other than T, or for non-simple multidimensional arrays,
3864 have not been addressed. They could almost certainly be handled the
3865 same way if anyone is motivated to do so.)
3866 * The changes in array handling break binary compatibility, so
3867 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3868 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3869 instead of (VALUES T T).
3870 * By following the instructions that Dan Barlow posted to sbcl-devel
3871 on 2 July 2000, I was able to enable primitive dynamic object
3872 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3873 functionality is not implemented (since it calls ld to resolve
3874 library references automatically, requiring RUN-PROGRAM for its
3875 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3876 try to resolve library references) is now supported.
3877 * The system now flushes the standard output streams when it terminates,
3878 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3879 them at several other probably-convenient times, e.g. in each pass of
3880 the toplevel read-eval-print loop, and after evaluating a form given
3881 as an "--eval" command-line option. (These changes were motivated by a
3882 discussion of stream flushing issues on cmucl-imp in August 2000.)
3883 * The source transform for TYPEP of array types no longer assumes
3884 that an array whose element type is a not-yet-defined type
3885 is implemented as an array of T, but instead punts, so that the
3886 type will be interpreted at runtime.
3887 * There is now some support for cross-compiling in make.sh: each of
3888 the phases of make.sh has its own script. (This should be transparent
3889 to people doing ordinary, non-cross-compile builds.)
3890 * Since my laptop doesn't have hundreds of megabytes of memory like
3891 my desktop machine, I became more motivated to do some items on
3892 my to-do list in order to reduce the size of the system a little:
3893 ** Arrange for various needed-only-at-cold-init things to be
3894 uninterned after cold init. To support this, those things have
3895 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3896 symbols with such names are now uninterned after cold init).
3897 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3898 fasl files for cold load.
3899 ** Remove the old compiler structure pooling code (which used to
3900 be conditional on the target feature :SB-ALLOC) completely.
3901 ** Redo the representation of some data in cold init to be more compact.
3902 (I also looked into supporting byte compiled code at bootstrap time,
3903 which would probably reduce the size of the system a lot, but that
3904 looked too complicated, so I punted for now.)
3905 * The maximum signal nesting depth in the src/runtime/ support code has
3906 been reduced from 4096 to 256. (I don't know any reason for the very
3907 large old value. If the new smaller value turns out to break something,
3908 I'll probably just bump it back up.)
3909 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3911 * Many, many bugs reported by Peter Van Eynde have been added to
3912 the BUGS list; some have even been fixed.
3913 * While enabling dynamic object file loading, I tried to make the
3914 code easier to understand, renaming various functions and variables
3915 with less ambiguous names, and changing some function calling
3916 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3917 * While trying to figure out how to do the OpenBSD port, I tried to
3918 clean up some of the code in src/runtime/. In particular, I dropped
3919 support for non-POSIX signal handling, added various comments,
3920 tweaked the code to reduce the number of compilation warnings, and
3921 renamed some files to increase consistency.
3922 * To support the new automatic configuration functionality in make.sh,
3923 the source file target-features.lisp-expr has been replaced with the
3924 source file base-target-features.lisp-expr and the machine-generated
3925 file local-target-features.lisp-expr.
3926 * fixed a stupid quoting error in make.sh so that using CMU CL
3927 "lisp -batch" as cross-compilation host works again
3929 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3931 * DESCRIBE no longer tries to call itself recursively to describe
3932 bound/fbound values, so that it no longer fails on symbols which are
3933 bound to themselves (like keywords, T, and NIL).
3934 * DESCRIBE now works on generic functions.
3935 * The printer now prints less-screwed-up representations of closures
3936 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3937 * A private symbol is used instead of the :EMPTY keyword previously
3938 used to mark empty slots in hash tables. Thus
3939 (DEFVAR *HT* (MAKE-HASH-TABLE))
3940 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3941 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3942 now does what ANSI says that it should. (You can still get
3943 similar noncompliant behavior if bang on the hash table
3944 implementation with all the symbols you get back from
3945 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3946 This breaks binary compatibility, since tests for equality to
3947 :EMPTY are wired into things like the macroexpansion of
3948 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3950 * There's now a minimal placeholder implementation for CL:STEP,
3951 as required by ANSI.
3952 * An obscure bug in the interaction of the normal compiler, the byte
3953 compiler, inlining, and structure predicates has been patched
3954 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3955 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3957 * Missing ordinary arguments in a macro call are now detected even
3958 when the macro lambda list contains &KEY or &REST.
3959 * The debugger no longer complains about encountering the top of the
3960 stack when you type "FRAME 0" to explicitly instruct it to go to
3961 the top of the stack. And it now prints the frame you request even
3962 if it's the current frame (instead of saying "You are here.").
3963 * As specified by ANSI, the system now always prints keywords
3964 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3966 * The default initial SIZE of HASH-TABLEs is now smaller.
3967 * Type information from CLOS class dispatch is now propagated
3968 into DEFMETHOD bodies, so that e.g.
3969 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3971 is now basically equivalent to
3972 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3973 (DECLARE (TYPE SINGLE-FLOAT X))
3975 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3976 operation, without having to do run-time type dispatch.
3977 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3978 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3979 * Since I don't seem to be making big file reorganizations very often
3980 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3981 is overflowing my ability to conveniently back them up), I've finally
3982 checked the system into CVS. (The CVS repository is on my home system,
3983 not at SourceForge -- putting it on SourceForge might come later.)
3984 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3985 high-level GC-NOTIFY-FOO functions send their output. (There's
3986 still very little control of where low-level verbose GC functions
3987 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3988 less than it used to -- the GC-NOTIFY-FOO functions are now under
3989 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3990 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3991 in only one place in the source code, and propagates it automatically
3992 everywhere that it's needed. Thus e.g. when I bump the version from
3993 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3994 * The C source files now include boilerplate legalese and documentation
3995 at the head of each file (just as the Lisp source files already did).
3996 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3997 to his page will be replaced with a link to his new CLiki service.
3999 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4001 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4002 * The build process now looks for GNU make under the default name "gmake",
4003 instead of "make" as it used to. If GNU make is not available as "gmake"
4004 on your system, you can change this default behavior by setting the
4005 GNUMAKE environment variable.
4006 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4007 can build without error under CMU CL.
4009 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4011 * There is now a partial SBCL user manual (with some new text and some
4012 text cribbed from the CMU CL manual).
4013 * The beginnings of a profiler have been added (starting with the
4014 CMU CL profiler and simplifying and cleaning up). Eventually the
4015 main interface should be through the TRACE macro, but for now,
4016 it's still accessed through vaguely CMU-CL-style functions and macros
4017 exported from the package SB-PROFILE.
4018 * Some problems left over from porting CMU CL to the new
4019 cross-compilation bootstrap process have been cleaned up:
4020 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4021 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4022 part when getting rid of the compiler *BACKEND* stuff.)
4023 ** The value of *NULL-TYPE* was screwed up, because it was
4024 being initialized before the type system knew the final
4025 definition of the 'NULL type. This screwed up several key
4026 optimizations in the compiler, causing inefficiency in all sorts
4027 of places. (I found it because I wanted to understand why
4028 GET-INTERNAL-RUN-TIME was consing.)
4029 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4030 PROCLAIM FTYPE information. Unfortunately this broke binary
4031 compatibility again, since now the forms output by DEFGENERIC
4032 to refer to functions which didn't exist in 0.6.3.
4033 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4034 can use the new (as of 0.6.3) transform for SOME into MAP into
4036 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4037 values to (INTEGER 0 1000000), so that the time code will no longer
4038 occasionally get blown up by Linux returning 1000000 microseconds
4039 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4040 its output conform to the ANSI spec. (Alas, this makes its output
4041 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4042 * A full call to MAP NIL with a single sequence argument no longer conses.
4043 * fixes to problems pointed out by Martin Atzmueller:
4044 * The manual page no longer talks about multiprocessing as though
4045 it were currently supported.
4046 * The ILISP support patches have been removed from the distribution,
4047 because as of version 5.10.1, ILISP now supports SBCL without us
4048 having to maintain patches.
4049 * added a modified version of Raymond Toy's recent CMU CL patch for
4050 EQUALP comparison of HASH-TABLE
4052 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4054 * The system still can't cross-compile itself with
4055 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4056 entails), but at least it can compile more of itself that way
4057 than it used to be able to, and various buglets which were uncovered
4058 by trying to cross-compile itself that way have now been fixed.
4059 * This release breaks binary compatibility again. This time
4060 at least I've incremented the FASL file format version to 2, so that the
4061 problem can be detected reliably instead of just causing weird errors.
4062 * various new style warnings:
4063 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4064 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4065 ** using the lexical binding of a variable named in the *FOO* style
4066 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4067 as specified by ANSI.
4068 * *RANDOM-STATE* is no longer automatically initialized from
4069 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4070 default behavior of the system is to repeat its behavior every time
4071 it's run. If you'd like to change this behavior, you can always
4072 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4073 old convention there was no comparably easy way to get the system to
4074 repeat its behavior every time it was run.
4075 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4076 FTYPE declarations has been removed, in favor of their ANSI
4077 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4078 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4079 the types of their sequence arguments can be determined at compile time.
4080 This is done through a new open code expansion for MAP which eliminates
4081 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4082 argument types can be determined at compile time.
4083 * The optimizer now transforms COERCE into an identity operation when it
4084 can prove that the coerced object is already of the correct type. (This
4085 can be a win for machine generated code, including the output of other
4086 optimization transforms, such as the MAP transform above.)
4087 * Credit information has been moved from source file headers into CREDITS.
4088 * Source file headers have been made more standard.
4089 * The CASE macro now compiles without complaining even when it has
4092 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4094 * (Note that the way that the PCL macroexpansions were rewritten
4095 to accommodate the change in DEFGENERIC below breaks binary
4096 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4097 not run under sbcl-0.6.2. Once we get out of alpha releases,
4098 i.e. hit release 1.0.0, we'll probably try to maintain binary
4099 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4100 and sbcl-1.4.4. Until then, however, it might be fairly common
4101 for maintenance releases to break binary compatibility.)
4102 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4104 * The reporting of unhandled signals has been changed to print some
4105 explanatory text as well as the report form. (Previously only
4106 the report form was printed.)
4107 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4108 it defines, so that the compiler no longer issues undefined function
4109 warnings for compiled-but-not-yet-loaded generic functions.
4110 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4111 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4112 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4113 section of the ANSI spec are supported.
4114 * The "" nickname for the "KEYWORD" package has been removed.
4115 The reader still handles symbol tokens which begin with a package marker
4116 as keywords, but it doesn't expose its mechanism for doing so in the
4117 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4118 * The system now issues STYLE-WARNINGs for contradictory TYPE
4119 proclamations. (Warnings for contradictory FTYPE proclamations would
4120 be nice too, but those can't be done usefully unless the type system
4121 is made smarter about FUNCTION types.)
4122 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4123 systematized, so that "*target-*.lisp is supposed to exist only on the
4124 target and imply that there's a related file which exists on the
4125 host, and *host-*.lisp is supposed to exist only on the host and imply
4126 that there's a related file which exists on the target. This involves a
4127 lot of renaming. Hopefully the acute confusion caused by the renaming
4128 will be justified by the reduction in chronic confusion..
4129 ** runtime-type.lisp -> early-target-type.lisp
4130 ** target-type.lisp -> late-target-type.lisp
4131 ** early-host-format.lisp -> early-format.lisp
4132 ** late-host-format.lisp -> late-format.lisp
4133 ** host-error.lisp -> misc-error.lisp
4134 ** early-error.lisp -> early-target-error.lisp
4135 ** late-error.lisp -> late-target-error.lisp
4136 ** host-defboot.lisp -> early-defboot.lisp
4137 ** code/misc.lisp -> code/target-misc.lisp
4138 ** code/host-misc.lisp -> code/misc.lisp
4139 ** code/numbers.lisp -> code/target-numbers.lisp
4140 ** code/early-numbers.lisp -> numbers.lisp
4141 ** early-host-type.lisp -> early-type.lisp
4142 ** late-host-type.lisp -> late-type.lisp
4143 ** host-typep.lisp -> typep.lisp
4144 ** load.lisp -> target-load.lisp
4145 ** host-load.lisp -> load.lisp
4146 ** host-disassem.lisp -> disassem.lisp
4147 ** host-insts.lisp -> insts.lisp
4148 ** byte-comp.lisp -> target-byte-comp.lisp
4149 ** host-byte-comp.lisp -> byte-comp.lisp
4150 ** host-signal.lisp -> signal.lisp
4151 ** host-defstruct.lisp -> defstruct.lisp
4152 ** late-target-type.lisp -> deftypes-for-target.lisp
4153 Furthermore, several other previously target-only files foo.lisp (e.g.
4154 hash-table.lisp and random.lisp) have been split into a target-and-host
4155 foo.lisp file and a target-only target-foo.lisp file, with their key type
4156 definitions in the target-and-host part, so that the cross-compiler will
4157 know more about target types.
4158 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4159 gone away. In their place are various *BACKEND-FOO* variables
4160 corresponding to the slots of the old structure.
4161 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4162 into the target SBCL has been fixed.
4163 * The chill.lisp system for loading cold code into a running SBCL
4165 * Support for the CMU CL "scavenger hook" extension has been removed.
4166 (It was undocumented and unused in the CMU CL sources that SBCL was
4167 derived from, and stale in sbcl-0.6.1.)
4168 * Various errors in the cross-compiler type system were detected
4169 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4170 (enabling various consistency checks). Many of them were fixed,
4171 but some hard problems remain, so the compiler is back to
4172 running without *TYPE-SYSTEM-INITIALIZED* for now.
4173 * As part of the cross-compiler type system cleanup, I implemented
4174 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4175 * I have started adding UNCROSS calls throughout the type system
4176 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4177 on cross-compiler input files will be able to go away, and various
4179 * CONSTANTP now returns true for quoted forms (as explicitly required
4182 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4184 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4185 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4186 (MAKE-LIST -1). In the longer term, it ought to become true
4187 that declarations are assertions even at SAFETY 1. For now, it's not
4188 quite true even at SAFETY 3, but it's at least more nearly true..
4189 (Note that this change seems to increases the size of the system by
4190 O(5%) and to decrease the speed of the compiler by 20% or more.)
4191 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4192 for the problem of printing dozens of lines of distracting information
4193 about low-level system machinery as part of the top stack frame
4194 on entry to the debugger when an undefined function was called.
4195 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4196 is not reset to COMMON-LISP-USER.
4197 * Compilation of stuff related to dyncount.lisp has been made conditional
4198 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4199 smaller. The various dyncount-related symbols have been moved into
4200 a new "SB-DYNCOUNT" package.
4201 * tty-inspect.lisp has been renamed to inspect.lisp.
4202 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4203 feature has gone away. (When we eventually port to other flavors of
4204 libc and/or Unix, we'll try to make the differences between flavors
4205 invisible at the user level.)
4206 * Various other *FEATURES* tags, and/or their associated conditionals,
4207 have been removed if obsolescent, or given better documentation, or
4208 sometimes given more-mnemonic names.
4210 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4212 * tidied up "make.sh" script
4213 * tidied up system directory structure
4214 * better "clean.sh" behavior
4215 * added doc/FOR-CMUCL-DEVELOPERS
4216 * many many small tweaks to output format, e.g. removing possibly-confusing
4217 trailing #\. character in DESCRIBE-INSTANCE
4218 * (EQUALP #\A 'A) no longer signals an error.
4219 * new hashing code, including EQUALP hashing
4220 * tidied up Lisp initialization and toplevel
4221 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4222 * command line argument processing
4223 * added POSIX-GETENV function to deal with Unix-ish environment variables
4224 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4225 terminating SBCL on EOF
4226 * non-verbose GC by default
4227 * There is no more "sbcl" shell script; the sbcl file is now the C
4228 runtime executable (just like CMU CL).
4229 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4231 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4232 symbol and package information as Lisp native symbol and package objects
4233 instead of strings naming symbols and strings naming packages. This way,
4234 whenever packages are renamed (as in warm init), debug information is
4235 transformed along with everything else.
4236 * tweaked the optimization policy declarations which control the building
4237 of SBCL itself. Now, among other things, the system no longer saves
4238 source location debugging information. (This helps two problems at once
4239 by reducing SBCL size and by keeping SBCL from trying to look for its
4240 sources -- which may not exist -- when reporting errors.)
4241 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4242 debugging and testing purposes
4243 * cleaned up printing, making the printer call PRINT-OBJECT for
4244 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4245 methods, giving nearly-ANSI behavior
4246 * converted almost all special variables to use *FOO* naming convention
4247 * deleted PARSE-TIME functionality, since it can be done portably
4248 * moved some files out of cold init into warm init
4249 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4251 * regularized formatting of source files
4252 * added an install.sh script
4253 * fixed ridiculous memory usage of cross-compiler by making
4254 compiler/alloc.lisp not try to do pooling unless it can hook
4255 itself into the GC of the cross-compilation host. Now the system
4256 builds nicely on my old laptop.
4257 * added :SB-ALLOC in target-features.lisp-expr
4258 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4259 was not implemented)
4260 * re-did condition handling and note reporting in the compiler. Notes
4261 are no longer handled by signalling conditions. Style warnings
4262 and warnings are handled more correctly and reported in such a way
4263 that it's easy to find one or the other in your output (so that you
4264 can e.g. figure out which of many problems caused COMPILE-FILE to
4266 * changed the severity of several compiler warnings from full WARNING
4267 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4268 compiler note reporting so that it doesn't use the condition system
4269 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4270 COMPILE-FILE command)
4271 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4272 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4273 no longer supported, since their implementation was deeply intertwingled
4274 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4275 * removed bogus "support" for compiler macros named (SETF FOO), and
4276 removed the compiler macro for SETF INFO (but only after making a fool
4277 of myself on the cmucl-imp mailing list by posting a bogus patch for
4278 DEFINE-COMPILER-MACRO..)
4279 * Compiled files containing forms which have side effects on the Lisp
4280 reader (such as DEFPACKAGE forms) are now handled more correctly.
4281 (Compiler queuing of top level lambdas has been suppressed by setting
4282 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4283 * deleted various currently-unused source files, e.g. gengc.lisp. They
4284 may be added back at some point e.g. when porting to other architectures,
4285 but until they are it's distracting to distribute them and to try to
4287 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4288 there were so many of them they're just distractions, and UNCROSS is
4289 known to be able to handle the current sources
4290 * moved PROFILE functionality into TRACE, so that it will be clear
4291 how the wrapping and unwrapping of functions when you profile them
4292 interacts with the wrapping and unwrapping of functions when you
4293 trace them. (Actually, the functionality isn't there yet, but at least
4294 the interface specification is there. Hopefully, the functionality will
4295 arrive with some future maintenance release.)
4296 * removed host-oops.lisp
4297 * changed signature of QUIT function to allow UNIX-CODE argument
4298 * fixed READ-SEQUENCE bug
4299 * tweaked verbose GC output so that it looks more like the progress
4300 output that ANSI specifies for functions like LOAD
4301 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4302 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob