1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
3 * MOP improvement: support for user-defined subclasses of
4 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
5 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
6 the expansion of DEFMETHOD, is responsible for generating a form
7 which creates a list of specializers when evaluated. Additional
8 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
9 debugging and introspective support.
10 * minor incompatible change: the (unsupported) spinlock interface
11 has changed: free spinlock now has the value NIL, and a held spinlock
12 has the owning thread as its value.
13 * enhancement: name of a socket-stream is now "a socket" instead of
15 * bug fix: the cache used by the CLOS to store precomputed effective
16 methods, slot offsets, and constant return values is now thread and
18 * bug fix: generic function dispatch function updating is now thread
19 and interrupt safe (in the sense that the known issues have been
22 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
23 * new contrib: sb-cover, an experimental code coverage tool, is included
25 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
27 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
28 produce more efficient code on x86-64 if the compiler can determine
29 that (+ INDEX <constant>) does not require a bounds check and FOO
30 has an element type at least 8 bits wide.
31 * enhancement: a new, experimental synchronous timeout facility is
32 provided. Refer to SB-SYS:WITH-DEADLINE for details.
33 * enhancement: when a symbol name conflict error arises, the
34 conflicting symbols are always printed with a package prefix.
35 (thanks to Kevin Reid)
36 * enhancement: stepping is now once again supported on the SPARC and
37 MIPS platforms. (It is also now more likely to work on CheneyGC
39 * enhancement: sb-sprof can now also track and report accurate call
41 * bug fixes: the treatment of non-standard subclasses of
42 SB-MOP:SPECIALIZER is more correct.
43 * incompatible change: PURIFY no longer copies the data from the
44 dynamic space into the static and read-only spaces on platforms
45 that use the generational garbage collector
46 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
48 * bug fix: GC race condition occasionally resulting in crashes with
49 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
51 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
53 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
54 a STYLE-WARNING to be signalled (regression from 1.0.4.)
55 * bug fix: an asynchronous interrupt could previously leave the
56 system running with GC inhibited.
57 * bug fix: a DECLARE form evaluated at top-level now causes an error
58 rather than silently (or verbosely) returning NIL.
59 * bug fix: trying to dissassemble functions compiled at high DEBUG could
60 result in a error being signalled due to source form lookup errors.
61 (reported by Peter Graves)
63 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
64 * incompatible change: removed writer methods for host-ent-name,
65 host-ent-addresses -- changing the values did not update the DNS
67 * minor incompatible change: changed experimental JOIN-THREAD interface
68 * documentation: the manual now lists reader and writer methods
69 in class slot documentation sections. (thanks to Richard M Kreuter)
70 * documentation: unwinding from asyncronous events has been
72 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
73 in multithreaded application code.
74 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
75 platforms. (thanks to James Anderson for the optimization hint)
76 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
78 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
79 variants no longer cons.
80 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
81 their NOT- variants no longer cons.
82 * optimization: EQUAL hash tables no longer use SXHASH for objects
83 of all data types, but instead use an EQL hash for types for which
84 EQUAL is the same as EQL.
85 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
86 are significantly faster.
87 * optimization: new STRING-OUTPUT-STREAM implementation is much
88 faster for large outputs and conses 30% less on average.
89 * enhancement: XREF information is now collected for references made
90 to global variables using SYMBOL-VALUE and a constant argument.
91 * enhancement: SIGINT now causes a specific condition
92 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
93 * bug fix: FILE-POSITION used to signal an error in some cases where
94 ANSI requires it to return NIL.
95 * bug fix: ADJUST-ARRAY is now interrupt-safe.
96 * bug fix: adding and removing fd-handlers is now interrupt-safe.
97 * bug fix: inlined calls to C now ensure 16byte stack alignment on
99 * bug fix: bad type declaration in the CLOS implementation has
100 been fixed. (reported by James Anderson)
101 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
102 (thanks to Lutz Euler)
103 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
104 required by the CLHS. (reported by Samium Gromoff)
105 * bug fix: dead unbound variable references now signal an error.
106 (reported by Marco Monteiro)
107 * bug fix: / with an unused value was being deleted in safe code.
108 (thanks to Marco Monteiro and Kevin Reid)
109 * bug fix: number of characters that can be written onto a single
110 line in a file is unlimited.
111 * bug fix: some GC deadlocks caused by asynchronous interrupts have
112 been fixed by inhibiting interrupts for when GC is disbled.
113 * bug fix: some interrupt safety issues with GETHASH, (SETF
114 GETHASH), CLRHASH and REMHASH have been fixed.
115 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
116 specifier no longer causes infinite recursion.
117 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
118 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
119 * bug fix: an &environment argument with the correct variable information
120 is passed to macros that are expanded during byte compilation.
121 (reported by Samium Gromoff)
122 * bug fix: a logic error was causing the hash function for bignums to
123 have excessive amounts of collisions. (reported by Faré Rideau)
124 * bug fix: modifying the contents of an array could change the return
125 value of SXHASH on that array, which is only allowed for strings
126 and bit vectors. (bug introduced in 0.9.16)
127 * improvement: the x86-64/darwin port now passes all tests (except
128 for the debugger tests) but should still be considered
129 experimental until this is fixed.
130 * improvement: a style-warning is signaled for CASE (etc) clauses with
131 duplicate keys. (thanks to Kevin Reid)
132 * improvement: macos/x86 and macos/x86-64 now use mach exception
133 handlers for dealing with illegal instructions (for trapping and
134 error handling) and memory protection violations (for GC).
136 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
137 * new platform: experimental support for x86-64/darwin (MacOS).
138 * incompatible change: the thread-safe (on most platforms) getaddrinfo
139 and getnameinfo sockets functions are used instead of gethostbyaddr
140 and gethostbyname, on platforms where the newer functions are available.
141 As a result, the ALIASES field of HOST-ENT will always be NIL on these
143 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
144 don't print the startup banner, but behave as if --noinform was passed
145 as a command line argument. (thanks to Kevin Reid)
146 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
147 * optimization: code using alien values with undeclared types is much faster.
148 * optimization: the compiler is now able to open code SEARCH in more cases.
149 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
150 * bug fix: using standardized COMMON-LISP special variables as loop
151 variables no longer signals bogus package lock violations. (reported
153 * bug fix: declaring local loop variables to be of a range-limited type
154 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
155 (reported by Andras Simon)
156 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
157 bugs remain on x86-64.)
158 * bug fix: the #= and ## reader macros now interact reasonably with
159 funcallable instances.
160 * bug fix: type-checks for function arguments were compiled using the
161 compiler policy settings of the wrong lexical environment (in 1.0.2
163 * bug fix: SHADOW accepts characters as string designators, as required
164 by the spec (thanks to Eric Marsden)
165 * bug fix: fixed GC safety issues when foreign functions are called with
166 non-base strings as arguments
167 * bug fix: more consistent error output for fatal-compiler-warnings, like
169 * bug fix: functions with proper names were showing up as NIL in
170 backtraces (reported by Edi Weitz, regression in 1.0.2)
172 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
173 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
174 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
175 produce more efficient code on the x86 if the compiler can determine
176 that (+ INDEX <constant>) does not require a bounds check and FOO
177 has an element type at least 8 bits wide.
178 * bug fix: references to undefined variables are handled the same way
179 in toplevel forms as in the normal compiler
180 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
182 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
183 (thanks to Jon Buller)
184 * improvement: faster compilation times for complex functions
185 * improvement: added readlink support to SB-POSIX (thanks to Richard
188 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
189 * improvement: experimental support for mach exception handling on
190 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
192 * improvement: support for GBK external format.
193 (thanks to Chun Tian (binghe))
194 * improvement: the debugger now displays variables that have been closed
195 over, in code compiled with (DEBUG 2) or higher
196 * improvement: support for executable cores on NetBSD (thanks to
198 * new feature: added a RESTART-FRAME debugger command
199 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
200 be used to provide an implementation for FILE-POSITION on Gray streams
201 (thanks to Eric Marsden)
202 * new feature: add syslog, openlog and closelog support to SB-POSIX
203 (thanks to Richard Kreuter)
204 * optimization: the function call overhead in code compiled with
205 a high DEBUG optimization setting is significantly
206 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
207 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
208 (thanks to Tony Martinez)
209 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
210 evaluated once (reported by Kevin Reid)
211 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
212 works again (thanks to Kilian Sprotte)
213 * bug fix: an error is signaled for tagbodies with duplicate tags
214 (thanks to Stephen Wilson)
215 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
216 * bug fix: Win32 port can now handle foreign code unwinding Lisp
217 stack frames from alien callbacks.
218 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
219 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
220 65k pages (thanks to David Woodhouse)
221 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
223 changes in sbcl-1.0.1 relative to sbcl-1.0:
224 * new platform: FreeBSD/x86-64, including support for threading.
225 * new feature: the compiler stores cross-referencing information
226 abount function calls (who-calls), macroexpansion (who-macroexpands)
227 and special variables (who-binds, who-sets, who-references) for code
228 compiled with (< SPACE 3). This information is available through the
229 sb-introspect contrib.
230 * new feature: users may subclass SEQUENCE, and have instances of
231 these classes interoperate with standard Common Lisp functions if
232 a number of methods are defined. (This feature is experimental
233 and the interface subject to change based on feedback from SBCL
234 users and the general community)
235 * improvement: sb-sprof traces call stacks to an arbitrary depth on
236 x86/x86-64, rather than the previous fixed depth of 8
237 * improvement: another pthread back-end of mutex "pthread-futex".
238 * bug fix: non-ascii command-line arguments are processed correctly
239 (thanks to Yaroslav Kavenchuk)
240 * bug fix: non-required arguments were not passed correctly when a method
241 defined using DEFMETHOD was called from a mop-generated method using
242 CALL-NEXT-METHOD (reported by Pascal Costanza)
243 * bug fix: recursion is now permitted in accessors through
244 SLOT-UNBOUND. (reported by Pascal Costanza)
245 * bug fix: an error was signaled at startup if the HOME environment
246 variable was defined, but had an empty value (reported by Peter Van Eynde)
247 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
248 signal a PROGRAM-ERROR, not a BUG.
249 * bug fix: SB-POSIX:READDIR works when built with large file support.
250 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
252 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
253 for better automatic indentation support. (thanks to Matt Pillsbury)
254 * optimization: loading generic functions no longer takes O(n^2) time,
255 proportional to the amount of methods in the generic function
256 (reported by Todd Sabin and Jeremy Brown)
257 * optimization: the FIND and POSITION family of sequence functions
258 are significantly faster on arrays whose element types have been
260 * improvements to the Windows port:
261 ** Intermittent heap corruption problems have been fixed. (thanks
262 to Alastair Bridgewater)
263 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
265 ** Lisp is able to unwind foreign exception frames from alien
266 callbacks. (thanks to Alastair Bridgewater)
268 changes in sbcl-1.0 relative to sbcl-0.9.18:
269 * improvement: experimental support for threading on FreeBSD/x86.
270 (thanks to NIIMI Satoshi)
271 * improvement: runtime option --dynamic-space-size can be used
272 to set the size of the dynamic space reserved on startup.
273 * improvement: floating point modes in effect are now saved in
274 core, and restored on startup.
275 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
276 startup, not time since first call to GET-INTERNAL-REAL-TIME.
277 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
278 threads are not running after *SAVE-HOOKS* have run.
279 * improvement: writes to CLOS instance slots are type-checked in code
280 compiled with (SAFETY 3)
281 * improvement: floating-point exception handling on FreeBSD (thanks to
283 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
284 (thanks to Zach Beane)
285 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
287 * improvement: added support for the Shift-JIS external format.
288 (contributed by NIIMI Satoshi)
289 * improvement: callbacks are supported on Linux/PPC. (thanks to
291 * bug fix: compiler bug triggered by a (non-standard) VALUES
292 declaration in a LET* was fixed. (reported by Kaersten Poeck)
293 * bug fix: file compiler no longer confuses validated and already
294 dumped structurres. (reported by Kaersten Poeck)
295 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
296 fill-pointer signals a type-error as required. (thanks to
298 * bug fix: disassemly of funcallable instances works.
299 * bug fix: single stepping on PPC.
300 * bug fix: fix thread-safety problems in the type system (generally
301 manifesting as nonsensical errors like "STRING is a bad type specifier
302 for sequences" or "The value 1 is not of type FIXNUM")
303 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
304 (reported by Josip Gracin)
305 * bug fix: an error is signaled for attempts to displace arrays with
306 incompatible element types (thanks to Mario Mommer)
307 * bug fix: more correct handling of wide characters in debug info
308 (bug reported by Attila Lendvai and fixed by Juho Snellman)
309 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
310 and don't cause extra consing
311 * optimization: MAP and MAP-INTO are significantly faster on vectors
312 whose elements types have been declared.
313 * Improvements to SB-SPROF:
314 ** Support for allocation profiling
315 ** Reduced profiling overhead, especially for long profiling runs
316 * Improvements to the Windows port:
317 ** floating point exceptions are now reported correctly.
318 ** stack exhaustion detection works partially.
319 ** more accurate GET-INTERNAL-REAL-TIME.
320 ** better RUN-PROGRAM behaviour in case of errors in the child
322 ** PROBE-FILE now simplifies pathnames correctly.
323 ** DIRECTORY now works correctly with :WILD-INFERIORS.
324 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
325 1st 1970 (although time timezone data will be unreliable).
326 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
328 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
329 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
330 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
332 * bug fix: two potential GC deadlocks affecting threaded builds.
333 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
334 returns T (reported by Anton Kazennikov)
335 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
336 with non-variable places
337 * bug fix: remove a race condition in the setting of
338 funcallable-instance functions, this should make threaded CLOS
339 code more stable against memory faults.
340 * bug fix: corruption of specials when unbinding is interrupted by an
341 asynchronous unwind (reported by Hannu Koivisto)
342 * improvement: the debugger will now also display local variables that
343 are only used once, for code compiled with a DEBUG optimization quality
346 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
347 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
348 * incompatible change: External-format support for FFI calls. The
349 SB-ALIEN:C-STRING no longer implies an ASCII
350 external-format. Instead the string is subject to external-format
351 conversion. Additionally return values of type SB-ALIEN:C-STRING
352 are by default of type (SIMPLE-ARRAY CHARACTER), not
353 SIMPLE-BASE-STRING. For an alien type that behaves like the old
354 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
355 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
356 * incompatible change: SB-EXT package no longer contains the
357 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
358 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
360 * incompatible change: the single-stepper is no longer available
361 on Alpha, Mips and Sparc platforms.
362 * minor incompatible change: the direct superclasses of
363 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
364 not (STANDARD-OBJECT FUNCTION). This makes the
365 class-precedence-lists of GENERIC-FUNCTION and
366 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
368 * minor incompatible change: the default stream external format on
369 non-unicode SBCL is no longer detected from the locale, but is always
371 * new feature: Add a version of evaluator that uses an interpreter instead
372 of the compiler. EVAL still uses the compiler by default, to switch it
373 to use the interpreter, set the value of the variable
374 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
375 * minor incompatible change: the single-stepper REPL has been merged
376 with the normal debugger (see the "Stepping" heading of the debugger help
377 for more details). The debugger command STEP will no longer switch
378 to the single-stepper REPL.
379 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
380 for a type now works.
381 * bug fix: loading of default sysinit file works. (thanks to Leonid
383 * bug fix: better detection of circularities in the file-compiler.
384 (reported by Marco Monteiro)
385 * bug fix: the CL pathname functions now work with files that have
386 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
387 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
388 systems with Unix98 pty semantics.
389 * bug fix: ASDF-INSTALL will now work with bsd tar.
390 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
392 * bug fix: timers expiring in dead threads no longer cause a
393 type-error (reported by Paul "Nonny Mouse").
394 * bug fix: thanks to more lightweight single-stepper instrumentation,
395 code compiled with (DEBUG 3) will compile and execute significantly faster,
396 and will have more accurate type-inferencing than before
397 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
398 whose bindings are modified
399 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
400 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
401 ** replace dummy implementations of CL:MACHINE-INSTANCE and
402 CL:SOFTWARE-VERSION with proper version
404 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
405 * feature: implemented the READER-METHOD-CLASS and
406 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
407 as specified by AMOP.
408 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
410 * optimization: faster LOGCOUNT implementation on x86 and x86-64
411 (thanks to Lutz Euler)
412 * optimization: hashing of general arrays and vectors has been
413 improved. (reported by Any Fingerhut)
414 * enhancement: SB-INTROSPECT is now able to find definitions of
415 profiled functions. (thanks to Troels Henriksen)
416 * enhancement: compiler-macro expansion applies now to FUNCALL forms
417 as well, allowing compiler-macros for SETF-functions to expand.
418 * enhancement: step-instrumentation no longer wraps known
419 single-value functions with multiple-value context, allowing
420 better type inference.
421 * fixed bug #337: use of MAKE-METHOD in method combination now works
422 even in the presence of user-defined method classes. (reported by
423 Bruno Haible and Pascal Costanza)
424 * fixed bug #339(c): if there are applicable methods not part of any
425 long-form method-combination group, call INVALID-METHOD-ERROR.
426 (reported by Bruno Haible)
427 * fixed bug #361: the :FUNCTION initarg in the protocol for
428 initialization of methods can now be used to override
429 internally-produced optimized functions. (reported by Bruno
431 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
432 system-provided lambda expression no longer cause warnings about
433 unbound #:|pv-table| symbols.
434 * bug fix: improved the handling of type declarations and the
435 detection of violations for keyword arguments with non-constant
437 * bug fix: potentially erronous calls to PATHNAME and
438 MERGE-PATHNAMES were being flushed in some cases.
439 (reported by Richard Kreuter)
440 * bug fix: compiled calls to TYPEP were mishandling obsolete
441 instances. (reported by James Bielman and Attila Lendvai)
442 * bug fix: format strings with ~> without matching ~< no longer
443 trigger an AVER, but signal an understandable error instead.
444 (reported by Antonio Martinez)
445 * bug fix: specifying an output-file without a pathname-type for
446 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
447 of input-file instead of "fasl". (reported by Robert Dodier)
448 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
449 (reported by James Y Knight).
450 * bug fix: compiler-macros-function did not consider the environment
451 argument for shadowing by local functions.
452 * bug fix: compiler-macros expansion was inhibited by local INLINE
454 * bug fix: inline expansions of known functions were subject to
455 step-instrumentation in high debug policies, leading to problems
457 * bug fix: compiler failed to differentiate between different CONS
459 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
460 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
461 * bug fix: #368: incorrect use of expressed vs. upgraded array
463 * bug fix: #306a: more precise unions of array types.
464 * thread-safety improvements:
465 ** CONDITION-WAIT could return early on Linux, if the thread was
466 interrupted and subsequently continued with SIGCONT.
467 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
469 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
470 * added support for the ucs-2 external format. (contributed by Ivan
472 * minor incompatible change: pretty printing of objects of type
473 (cons symbol) is, in the default pprint-dispatch-table, now
474 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
476 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
477 called later in a class's lifetime, possibly as late as when the
478 first instance of the class is created. Previously,
479 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
480 class became finalizeable.
481 * fixed bug: FILE-POSITION sometimes returned inconsistent results
482 for multibyte external-format streams. (thanks to "vbzoli")
483 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
484 with :ALLOCATION :CLASS inherited from superclasses of the
486 * fixed bug: anonymous classes can now be created using the :NAME
487 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
488 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
489 * fixed bug: core-files saved with :EXECUTABLE T can again be
490 executed when SBCL_HOME isn't set. (reported by James Knight)
491 * fixed bug: toplevel LOCALLY forms with declarations could
492 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
493 * fixed bug: printing from several different threads using different
494 values of *print-case* could cause invalid output, due to
495 some internal special variables of the printer not being bound
496 thread-locally (reported by Max Mikhanosha)
497 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
498 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
499 (reported by Pascal Costanza)
500 * fixed bug: FINALIZE-INHERITANCE is called from
501 REINITIALIZE-INSTANCE on classes when the class has previously
502 been finalized, as required by AMOP.
503 * minor code generation optimizations:
504 ** better register allocation in CLOS dispatching functions
505 ** overflow detection when coercing signed bytes to fixnums on x86-64
506 ** is now implemented with one IMUL instruction instead of three shifts
507 ** more efficient bit-vector access on x86 and x86-64
508 ** more efficient access to raw structure slots on x86 and x86-64
509 * fixed some bugs revealed by Paul Dietz' test suite:
510 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
513 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
514 * feature: thread support on Solaris/x86, and experimental thread support
516 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
518 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
519 default method for SLOT-UNBOUND.
520 * minor incompatible change: prevent the user from specializing the
521 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
522 somewhat counter to the intent of the protocol, I (CSR) think, and
523 additionally it just doesn't work in SBCL as currently
524 implemented, thanks to optimizations (that are always valid for
525 the other three SLOT-VALUEish functions, but not for the setter).
526 * minor incompatible change: the :SB-LDB feature is now enabled by
527 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
528 the low-level debugger.
529 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
530 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
531 down to an SB-SYS:FD-STREAM, allowing direct communication in
532 more cases. Temporary files and pipes are still used as a fallback
534 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
535 on the non-reentrant C functions localtime(3) and gmtime(3).
536 * bug fix: global optimization policy was not visible in LOCALLY and
538 * bug fix: class objects can be used as specializers in methods.
539 (reported by Pascal Costanza)
540 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
541 with a type error. (reported by blitz_ on #lisp)
542 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
543 failed with a type error. (reported by Pascal Bourguignon)
544 * bug fix: merging pathnames against defaults with :DIRECTORY
545 starting with '(:RELATIVE :BACK) should preserve the :BACK.
546 (reported by James Y Knight)
547 * bug fix: saving large (>2GB) cores on x86-64 now works
548 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
549 constant index and a (simple-array (signed-byte 32)) array
550 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
551 error (patch by Robert J. Macomber)
552 * feature: TIME now displays the amount of run-time spent in GC
553 * bug fix: The debugger now does a better job of respecting
554 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
555 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
556 workaround for bug 403.)
557 * bug fix: floating point exception handling now works on FreeBSD
558 (thanks to NIIMI Satoshi)
559 * fixed some bugs revealed by Paul Dietz' test suite:
560 ** MISC.641: LET-conversion were not supposed to work in late
562 * improvements to the Win32/x86 port:
563 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
564 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
566 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
567 * new feature: source path information is generated for macro-expansion
568 errors for use in IDE's like Slime (thanks to Helmut Eller)
569 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
570 * bug fix: compiler does not loop forever on an invalid type in
572 * improvement: compilation of most CLOS applications is significantly
574 * optimization: added a limited bytecode compiler for simple toplevel
575 forms, speeding up compilation and FASL loading
576 * bug fix: the statistical profiler now properly distinguishes anonymous
579 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
580 * minor incompatible change: in sbcl-0.9.11 (but not earlier
581 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
582 into subclasses of the finalized class. Now user calls to
583 FINALIZE-INHERITANCE finalize just the one class, and calls by the
584 system return before any subclasses are finalized.
585 * minor incompatible change: The reader no longer ignores errors
586 regarding non-existent packages in #+ and #- feature tests.
587 * new feature: command line options --no-sysinit, --no-userinit to
588 inhibit loading the corresponding init files
589 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
590 for sending data through UDP sockets (thanks to François-René Rideau)
591 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
592 error is signalled instead (thanks to François-René Rideau)
593 * minor incompatible change: Error signalling behaviour of lexical
594 operations violating package locks has changed slightly. Refer to
595 documentation on package locks for details.
596 * enhancement: EVAL can process IF-expressions without resorting to the
598 * bug fix: Some operations involving SIN, COS, or TAN weren't being
599 constant-folded properly due to missing out-of-line functions.
600 (reported by Mika Pihlajamäki)
601 * bug fix: LISTEN sometimes returned T even in cases where no data was
602 immediately available from the stream
603 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
604 were reversed. (reported by Levente Mészáros)
605 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
606 when given a vector displaced to an adjustable array. (reported
608 * bug fix: derivation of float boundaries from numbers outside the
609 appropriate float range (reported by John Wiseman)
610 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
612 * bug fix: precomputing cache entries for generic functions with
613 some subclasses of specializers as yet invalid does not attempt to
614 fill a cache line with a negative offset. (reported by Levente
616 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
617 allow condition class objects as arguments to DOCUMENTATION and
618 (SETF DOCUMENTATION); only find and set documentation for
619 structure names for the STRUCTURE doc-type. (suggested by Gary
621 * improvements to the Win32/x86 port:
622 ** (user-homedir-pathname) and default initialization file
623 locations now know about the user's "Documents and Settings"
624 directory (thanks to Yaroslav Kavenchuk)
625 ** run-program is implemented (thanks to Mike Thomas)
626 ** sockets support (thanks to Timothy Ritchey)
627 ** better backtrace support (thanks to Alastair Bridgewater)
628 ** sb-grovel supported
629 ** asdf-install and sb-posix work somewhat
630 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
631 * improvements to the Solaris/x86 port:
632 ** works on Solaris 11/Solaris Express
633 ** floating-point exception handling support
634 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
635 * fixed some bugs revealed by Paul Dietz' test suite:
636 ** REMOVE-METHOD returns its generic function argument even when
637 the method is not one of the generic functions' methods.
638 ** objects declared as MEMBER types can be admissible arguments to
640 ** printing characters should simply be printed by the FORMAT ~:C
642 ** compiler failure when compiling functions with hairy constant
643 defaults for optional parameters.
644 ** compiler produces wrong code when MAYBE-INLINE-expanding a
645 function, which is already optimized.
647 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
648 * new platform: experimental support for SBCL x86/Darwin, including
649 MacOS X 10.4.5 on Intel.
650 * new feature: Unicode character names are now known to the system
651 (through CHAR-NAME and NAME-CHAR).
652 * new feature: the filesystem location of SBCL's core file is
653 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
654 * minor incompatible change: the contrib modules SB-POSIX and
655 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
656 this change is to make it easier to distribute
657 location-independent binaries.
658 * bug fix: as implied by AMOP, standardized classes no longer have
659 slots named by external symbols of public packages. (reported by
661 * optimization: calling functions via constant symbols -- as in
662 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
663 via the function object as in (FUNCALL #'FOO).
664 * optimization: on x86 and x86-64, the calling convention for the
665 case of dealing with an unknown number of values has been altered
666 to be friendlier to the prediction heuristics implemented,
667 particularly with reference to CALL and RET pairing. (thanks to
668 Alastair Bridgewater)
669 * optimization: on x86, the code for access to thread-local symbol
670 values and binding/unbinding of thread-local symbols is smaller.
671 (thanks to Alastair Bridgewater)
672 * enhancement: CONSTANTP is now able to determine constantness of
673 more complex forms, including calls to constant-foldable standardized
674 functions and some special forms beyond QUOTE.
675 * fixed bug: occasional GC crashes on Solaris/x86
676 * optimization: x86-64 supports stack allocation of results of simple
677 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
678 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
679 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
680 (thanks to James Knight)
681 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
682 by the standard (thanks to Stephen Compall)
684 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
685 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
686 be used for bundling the runtime and the core file into one
687 executable binary. This feature is not currently supported on all SBCL
688 platforms. (thanks to James Bielman and NIIMI Satoshi)
689 * new feature: a generational or ephemeral garbage collector is now
690 the default on the PowerPC platform (both Linux and Darwin). The
691 old Cheney (stop and copy) collector is a build-time option.
692 (thanks to Cyrus Harmon, after Raymond Toy)
693 * minor incompatible change: the method by which SBCL finds its
694 contributed modules has changed; it no longer relies on symbolic
695 links from an $SBCL_HOME/systems directory, but searches directly
696 in the subdirectories of $SBCL_HOME.
697 * enhancement: the dynamic heap size on the Linux/PPC platform is
698 markedly larger, even using the older Cheney garbage collector.
699 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
700 certain complicated string types. (reported by Gary King)
701 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
702 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
703 * fixed bug: DOCUMENTATION on structures no longer signals an
704 error if no structure type of the right name exists. (reported by
706 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
707 applicable non-standard primary or :AROUND method on
708 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
709 no longer cause unbound variable errors. (reported by Kevin Reid)
710 * fixed bug: metacircle resolution in cases where methods have slots
711 added before the slots from STANDARD-METHOD. (reported by Jean
713 * fixed bug: the Power PC assembly code for calling into lisp
714 sometimes computed its return address wrongly (depending on the
715 15th bit of the address, fixed at link time). This used to
716 manifest itself in a segmentation violation while building PCL.
717 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
718 many others over the years)
719 * fixed bug: ROOM no longer reports silly numbers for stack usage if
720 the stack pointer (treated as a signed integer) is negative.
721 (thanks to Peter van Eynde)
723 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
724 * new platform: experimental support for the Windows operating
725 system has been added. (thanks to Alastair Bridgewater)
726 * fixed several bugs in and robustified the PPC FFI (including
727 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
728 * bug fix: allow non-simple string symbol names (reported by Paul
730 * bug fix: interrupt handling on NetBSD (thanks to Richard M
732 * bug fix: saving a core corrupted callbacks on x86/x86-64
733 * bug fix: closed a loophole in metacircularity detection and
734 grounding in the PCL implementation of CLOS.
735 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
736 (reported by tomppa on #lisp)
737 * bug fix: clear the direction flag on Lisp -> C transitions, as
738 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
739 (reported by Andrej Grozin and Hendrik Maryns)
740 * optimization: major improvements to GC efficiency on GENCGC platforms
741 * optimization: faster implementation of EQUAL
742 * optimization: emit more efficient opcodes for some common
743 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
744 * optimization: several other minor code-generation improvements on x86-64
745 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
747 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
748 * minor incompatible change: (SETF CLASS-NAME) and (SETF
749 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
750 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
751 by ANSI as a generic function, and for consistency (SETF
752 GENERIC-FUNCTION-NAME) is treated likewise.
753 * fixed bug #233.b: make constraint propagation notice when a variable
754 value is changed after it is referenced but before it is used
755 * fixed bug #296: no more arbitrary behaviour from filenames with
756 odd characters as --load arguments. (reported by Adam Warner)
757 * fixed bug #390: :CHARACTER-SET pathname components now work as
758 expected. (reported by Tim Daly Jr)
759 * fixed bug #391: complicated :TYPE intersections in slot
760 definitions no longer cause an error in PCL internals.
761 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
762 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
763 returning the number of octets which would be written to the
764 file-stream. (thanks to Robert J. Macomber)
765 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
766 arguments to the internal stream functions. (thanks to David
768 * bug fix: stack exhaustion is now handled on the x86/SunOS
769 platform. (thanks to Daisuke Homma)
770 * bug fix: more accurate ROOM results on GENCGC platforms
771 * optimization: improved type inference for arithmetic-for
772 index variables in LOOP
773 * optimization: faster floating-point SQRT on x86-64
774 * fixed some bugs revealed by Paul Dietz' test suite:
775 ** DOCUMENTATION returns NIL instead of "" for method combinations
776 that don't have a docstring
778 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
779 * minor incompatible change: (SETF CLASS-NAME) and (SETF
780 GENERIC-FUNCTION-NAME) are no longer generic functions, and
781 therefore are not customizeable by user code (as seems to be at
782 least permitted and maybe required by AMOP). As a consolation,
783 however, the SBCL implementation of these functions now calls
784 REINITIALIZE-INSTANCE as specified by AMOP.
785 * bug fix: slot-definition documentation is propagated as per ANSI
786 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
787 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
788 argument, as do eslotd initialization methods. (from Pascal
789 Costanza's "Closer" project)
790 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
791 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
793 * bug fix: it is now possible to have more than one subclass of
794 STANDARD-GENERIC-FUNCTION without causing stack overflow.
795 (reported by Bruno Haible, Pascal Costanza and others)
796 * bug fix: the dependent update protocol now works for generic
797 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
799 * bug fix: condition-class instances corresponding to
800 DEFINE-CONDITION forms are now created eagerly. (reported by
801 Kalle Olavi Niemitalo on comp.lang.lisp)
802 * bug fix: floating point printing is more accurate in some
803 circumstances. (thanks to Simon Alexander)
804 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
805 merged with *DEFAULT-PATHNAME-DEFAULTS*.
806 * bug fix: callbacks on OS X now preserve stack-alignment.
807 (thanks to Cyrus Harmon)
808 * enhancement: the x86-64 disassembler is much better at
809 disassembling SSE instructions. (thanks to Lutz Euler)
810 * enhancement: CHECK-TYPE on a variable now allows type-inference
811 (thanks to Vincent Arkesteijn)
812 * optimization: improved performance of EUC-JP external format.
813 (thanks to NIIMI Satoshi)
814 * optimization: performance improvements to IO on file streams of
815 :ELEMENT-TYPE CHARACTER
816 * optimization: much faster memory allocation on x86-64
817 * optimization: faster garbage collections (latency and throughput)
819 * optimization: faster bignum subtraction and division on x86 and x86-64
820 (thanks to Lutz Euler)
821 * optimization: more accurate type-inference for loops that use a
822 floating point index variable or a negative step.
824 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
825 * bug fix: add a workaround to SBCL looping infinitely at startup on
826 Linux kernels with apparently buggy implementations of personality().
827 (thanks to Svein Ove Aas)
828 * bug fix: Unicode symbols are correctly printed in LDB backtraces
829 (thanks to David Lichteblau)
830 * bug fix: local bindings shadow global symbol macros for
831 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
832 * optimization: non-open coded uses of numeric comparison operators
833 (e.g. >) no longer cons when called with more than one parameter
834 on platforms supporting dynamic-extent allocation.
835 * enhancement: saving cores with foreign code loaded is now supported
836 on MIPS/Linux in addition to the previously supported platforms.
837 * bug fix: FIND-RESTART now tests for activity, not applicability when given
838 a restart object as identifier. (reported by Helmut Eller for CMUCL)
839 * bug fix: division by zero in sb-sprof when no samples were collected
840 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
841 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
842 (thanks to Alexander Kjeldaas)
843 * new feature: ignoring the return values of destructive functions
844 like NREVERSE signals a compile-time style-warning.
845 (thanks to Kevin Reid)
846 * new feature: the alignment of alien structure fields can be
847 explicitly specified. (thanks to Cyrus Harmon)
848 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
849 and dump core on SIGQUIT
851 ** incompatible change: threads do not inherit values of specials
852 from their parents (see manual)
853 ** bug fix: threads stacks belonging to dead threads are freed by the
854 next exiting thread, no need to gc to collect thread stacks anymore
855 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
856 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
857 ** bug fix: dynamic variable and thread start related gc lossage
858 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
860 * fixed some bugs revealed by Paul Dietz' test suite:
861 ** SUBTYPEP is slightly more accurate on heinously complicated
862 CONS types where some of the members have uncertain (in the
863 NIL, NIL sense) type relationships to each other.
865 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
866 no longer purified unless :PURIFY T is explicitly specified.
867 ** Non-purified cores are significantly smaller than before
869 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
870 * new feature: timers based on Zach Beane's excellent timer package
871 * added support for the following external formats: koi8-u,
872 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
873 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
874 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
875 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
876 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
877 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
878 * incompatible change: a threaded SBCL will no longer revert to
879 non-threaded mode on non-NPTL systems, but refuse to start entirely.
880 * bug fix: interrupts are disabled until startup is complete; no
881 more sigsegvs when receiving a signal to soon
882 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
884 * bug fix: add a workaround for the memory randomization features in
885 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
886 workaround will only be in effect on systems with the proc filesystem
888 * bug fix: printing objects of type HASH-TABLE signals a
889 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
891 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
892 have been running for over 50 days. (reported by Gilbert Baumann)
893 * bug fix: the logic for getting names of functions gets less
894 confused when confronded with alternate-metaclass
895 funcallable-instances. (reported by Cyrus Harmon)
896 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
897 consistent, even on internal alternate-metaclass objects.
898 * bug fix: flush closure information collected by physical
899 environment analysis prepass before the main pass. (bug reported
901 * bug fix: compiler pack phase does not modify a hashtable, which is
902 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
903 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
904 may be expanded. (reported by Alexander <malishev> on c.l.l)
905 * bug fix: no more style-warnings when compiling code using LAST
906 under high-SPEED low-SPACE settings. (reported by David Wallin
908 * fixed some incompatibilities between SBCL's MOP and the MOP
910 ** the METAOBJECT class is now implemented;
911 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
912 STANDARD-OBJECT, as required;
913 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
914 now compatible (as required by VALIDATE-SUPERCLASS); there
915 remains a consistency requirement over the presence or absence
916 of the FUNCTION class in the superclass list of finalized
917 classes; see the manual for more details;
918 ** the :AROUND method for COMPUTE-SLOTS on
919 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
920 requested slot ordering.
922 ** bug fix: parent thread now can be gc'ed even with a live
924 ** bug fix: binding a special with PROGV to no value is not
925 the same as that symbol not having been bound (thanks to
927 ** bug fix: binding specials is thread safe (thanks to
929 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
930 and SB-SPROF do not die with 'no handler for signal XX in
931 interrupt_handle_now(..)' anymore
932 ** bug fix: WITH-TIMEOUT works with multiple running threads
933 * fixed some bugs revealed by Paul Dietz' test suite:
934 ** ENSURE-GENERIC-FUNCTION should take a method class object for
935 the :method-class keyword argument.
937 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
938 * new port: the Solaris operating system on x86 processors is now
939 mostly supported, though some rough edges in the environment
940 remain. (thanks to Daisuke Homma)
941 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
942 space for its heap. (thanks to Thiemo Seufer)
943 * optimization: SBCL on MIPS platforms now supports dynamic-extent
944 closures. (thanks to Thiemo Seufer)
945 * minor incompatible change: eof selects abort in the debugger.
946 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
947 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
949 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
950 deprecated, and will go away in a future revision of SBCL.
951 * minor incompatible change: GC-ON and GC-OFF are no longer
952 implemented with a counter, it does not matter how many times gc
953 is switched on or off
954 * bug fix: discriminating functions for generic function classes
955 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
956 make invalid assumptions about method precedence order. (reported
958 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
959 platforms. (thanks to Thiemo Seufer)
960 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
961 * bug fix: correct pathname printing with printer escaping is on.
962 (thanks to Kevin Reid)
963 * bug fix: complex VOP definitions in "user-space" no longer trigger
964 package locks. (reported by Zach Beane)
965 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
966 for funcallable-instance functions, and is no different from
968 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
969 characters in the current readtable. (reported by Nicholas Neuss)
970 * bug fix: SUBTYPEP on various CONS types returns more a more
971 accurate acknowledgment of its certainty.
972 * optimizations: REMOVE-DUPLICATES now runs in linear time on
973 lists in some cases. This partially fixes bug 384.
974 * flush all standard streams before prompting in the REPL and the
976 * bug fix: signal handling and triggering gc do not conflict
977 directly anymore, in particular a high frequency sb-sprof does
978 not prevent gc from running
979 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
980 approximation for timezone and DST information between the
981 universal time epoch and the smallest negative 32-bit time_t.
982 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
983 year 1899 so long as the encoded time is non-negative.
984 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
985 the correct answer for non-constant rotate amounts when performing
986 an inline 32-bit rotation.
988 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
989 there is only one thread in the session
990 ** bug fix: memory leak for streams created in one thread and
991 written to in another
992 ** bug fix: lockup when compiled with gcc4
993 ** bug fix: race that allows the gc to be triggered when gc is
995 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
996 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
998 ** bug fix: finished threads are gc'ed properly
999 * fixed some bugs revealed by Paul Dietz' test suite:
1000 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1001 has arguments with a different set of applicable methods from
1002 the orignal arguments.
1003 ** The type error thrown by MAP now has a correct expected-type
1005 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1006 name a compiled function.
1007 ** Three MISC tests where a large bignum was improperly coerced to
1008 a float that couldn't represent that bignum during type
1009 derivation were fixed.
1010 ** SUBTYPEP can now handle types involving the negation of a
1011 list-form FUNCTION type.
1012 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1013 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1014 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1016 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1017 * New feature: Experimental support for bivalent streams: streams
1018 opened with :element-type :default now allow character and binary
1019 (unsigned-byte 8) I/O
1020 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1021 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1023 * bug fix: no more highly sporadic "couldn't check whether ~S is
1024 readable" when reading a stream and an interrupt hits in the middle
1025 of a select system call
1026 * compiler better recognizes complex arrays (reported by Eduardo
1028 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1029 for variables on SBCL built with linkage-tables. (reported by Luis
1031 * various error reporting improvements.
1032 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1033 (Thanks to Thiemo Seufer)
1034 * bug fix: nested reader invokations work correctly
1035 * bug fix: it is possible to have simultaneous references to foreign
1036 code and foreign data with the same name.
1038 ** added x86-64 support
1039 ** incompatible change: the threading api now works with thread
1040 objects instead of thread ids
1041 ** bug fix: threads are protected from signals and interruption when
1042 starting up or going down
1043 ** bug fix: a race where an exiting thread could lose its stack to gc
1044 ** bug fix: don't halt on infinite error in threads if possible
1045 ** fixed numerous gc deadlocks introduced in the pthread merge
1046 ** bug fix: fixed thread safety issues in read and print
1047 ** bug fix: debugger doesn't hang on session lock if interrupted at
1048 an inappropriate moment
1049 ** bug fix: run-program is now thread safe(r)
1050 ** bug fix: inner with-recursive-lock no longer releases the mutex
1051 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1052 * fixed some bugs revealed by Paul Dietz' test suite:
1053 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1054 EXPECTED-TYPE slots filled.
1055 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1057 ** MAKE-CONDITION accepts classes as type-designators.
1058 ** COMPILE may never return NIL.
1059 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1060 range before calling Unix time functions
1062 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1063 * numerous signal handling fixes to increase stability
1064 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1065 * minor incompatible change: we now correctly canonize default
1066 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1067 as per the MOP, rather than the historical (INITARG INITFUNCTION
1068 INITFORM). (reported by Bruno Haible)
1069 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1070 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1071 for more information.
1072 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1073 pathname is a directory pathname.
1074 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1075 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1077 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1078 :METACLASS argument in addition to a class name. (reported by
1079 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1080 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1081 (reported by Bruno Haible; patch from Gerd Moellmann)
1082 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1084 * bug fix: more cleanups to the floating point exception handling on
1085 x86-64 (thanks to James Knight)
1086 * bug fix: the compiler does not try to constant fold calls of
1087 COERCE and BIT-* functions when they return freshly constructed
1088 objects. (reported by Nikodemus Siivola and Paul Dietz)
1089 * optimization: improved the MIPS versions of generic (in the
1090 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1091 * optimization: direct conversion of (unsigned-byte 32) to floats on
1092 the PowerPC platform.
1093 * optimization: structure instances with raw slots now use less
1094 memory, and probably show better memory locality. (thanks to
1096 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1097 are treated as requests for stack allocation on the x86-64,
1098 Alpha-32, PPC and SPARC platforms.
1099 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1100 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1101 (thanks to Tony Martinez)
1103 ** gcing a dead thread can no longer lead to lockups
1104 ** threads block signals until they are set up properly
1105 ** errno is no longer shared by threads
1106 ** interrupt-thread restores the eflags register on x86
1107 ** fixed some lockups due to gc/thread interaction
1108 * fixed some bugs revealed by Paul Dietz' test suite:
1109 ** invalid dotted lists no longer raise a read error when
1110 *READ-SUPPRESS* is T
1111 ** signal an error if a symbol that names a declaration is used
1112 as the name of a type, or vice versa
1113 ** allow using the (declare (typespec var*)) abbreviation for
1114 (declare (type typespec var*)) with all type specifiers
1115 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1116 ** always return NIL from PPRINT-POP when OBJECT is NIL
1117 ** don't signal errors when pretty-printing malformed LABELS,
1118 FLET or MACROLET forms
1119 ** declarations in a DOLIST body are also in scope for the
1121 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1123 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1126 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1127 * fixed cross-compiler leakages that prevented building a 32-bit
1128 target with a 64-bit host compiler.
1129 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1130 opened with :IF-EXISTS :OVERWRITE.
1131 * fixed bug 281: error for an invalid qualifier in a short-form method
1132 combination method is not signalled until the faulty method is called.
1133 * bug fix: iteration variable type inferrer failed to deal with open
1134 intervals. (reported by Alan Shields)
1135 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1136 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1137 or more runtime options were provided to the sbcl binary.
1138 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1139 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1140 * optimization: there's now a fast-path for fixnum arguments in the
1141 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1142 * optimization: the code generated on x86-64 is more compact thanks
1143 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1144 * bug fix: floating-point exception handling now partly works on
1145 x86-64. (Thanks to James Knight)
1146 * improvement to the MIPS backend from Thiemo Seufer: C-style
1147 64-bit long long arguments and return values to alien functions
1149 * fixed some bugs revealed by Paul Dietz' test suite:
1150 ** the type-error signalled from WARN has a filled-in DATUM slot.
1151 ** the type-error required when a stream is not associated with
1152 a file has the stream as its datum.
1153 ** type-errors on single-floats on x86-64 no longer have
1154 :INVALID-OBJECT as the datum
1155 ** the type-errors signalled for invalid function names now have
1156 a correct expected type
1157 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1158 for structures defined with a :TYPE. Documentation strings for
1159 typed structures are no longer immediately discarded
1160 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1161 broadcast streams and synonym streams. FILE-LENGTH now also works
1162 on broadcast streams.
1164 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1165 * incompatible change: the --noprogrammer option, deprecated since
1166 version 0.7.5, has been removed. Please use the equivalent
1167 --disable-debugger option instead.
1168 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1170 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1171 inoperational for a while now) has been completely removed.
1172 * null lexical environments are now printed as #<NULL-LEXENV>,
1173 significantly reducing the amount of clutter in typical backtraces.
1174 * documentation on weak pointers, finalization, and after GC hooks
1175 has been added to the manual.
1176 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1177 as other specialized array types, is much faster. SUBSEQ and
1178 COPY-SEQ on such arrays have also been sped up.
1179 * optimization: EQL is now more efficient when at least other argument
1180 is known to be of type (OR FIXNUM (NOT NUMBER)).
1181 * fixed bug: the runtime is now less vulnerable to changes in the
1182 size of the SBCL object on OS X, and virtual memory is reserved for
1183 all spaces that need to be at a fixed address.
1184 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1185 * fixed bug: finalizers and after GC hooks that cause consing are now
1187 * fixed bug: compiler error messages and summaries are now printed to
1188 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1189 * fixed inference of the upper bound of an iteration variable.
1190 (reported by Rajat Datta).
1191 * fixed bug 211e: calling local functions with duplicated constant
1192 keyword argument no longer causes a bogus style warning about an
1194 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1195 ftype declaration to be disregarded. (reported by Dave Roberts)
1196 * fixed bug 373: caused by erronous compilation of references to alien
1197 variables in the runtime on ppc/darwin.
1198 * fixed bug 376: CONJUGATE type deriver.
1199 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1200 (reported by Baughn on #lisp)
1201 * a cross-compiler bug on non-x86 platforms has been identified and
1202 fixed. (thanks to Bruno Haible)
1203 * improvements to the MIPS runtime code for increased stability.
1204 (thanks to Thiemo Seufer)
1205 * increased the maximimum compact environment size to allow
1206 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1207 * partial workaround for bug 135: don't name the function we're
1208 calling for hairy cases of EVAL, so as not to accumulate one environment
1209 entry per call to EVAL. (reported by Kevin Reid)
1210 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1211 * optimization: type testing for non-vector arrays should be faster.
1212 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1214 * bug fix: setting 31st element of a bit vector to zero did not work
1216 * bug fix: redefining a class definition which failed due to a
1217 previous accessor / function clash now works (but see BUGS entry
1218 #380 for more problems in this area). (thanks to Zach Beane)
1219 * the long form of DEFINE-METHOD-COMBINATION disables method group
1220 checking when given a single method group with pattern *. (thanks
1222 * on x86 compiler supports stack allocation of results of simple
1223 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1224 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1225 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1226 * fixed some bugs related to Unicode integration:
1227 ** the restarts for recovering from input and output encoding
1228 errors only appear when there is in fact such an error to
1230 * fixed some bugs revealed by Paul Dietz' test suite:
1231 ** MISC.549 and similar: late transformation of unsafe type
1232 assertions into derived types caused unexpected code
1234 ** SCALE-FLOAT type deriver is less wrong.
1235 ** type derivers for EXP, LOG and similar functions compute result
1236 types for complex arguments better.
1237 ** (MISC.563) CONJUGATE type deriver works for very restricted
1239 ** out-of-line type testers for character strings are available.
1240 ** EQUAL compiler transform understands specialness of objects
1242 ** accessing double-floats stored on the stack now works on x86-64.
1243 ** debugger internals could sometimes create invalid lispobjs,
1244 resulting in GC crashes.
1245 ** MISC.548: type check weakening can convert required type into
1247 ** initialization forms for bindings are not in scope of free special
1250 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1251 * incompatible change: thread support for non-NPTL systems has
1252 been removed - locking is buggy and unreliable. A threaded
1253 SBCL build will now warn at startup and refuse to create
1254 new threads, unless futex support is detected in the kernel
1255 * incompatible change: the top level REPL now has only an ABORT
1256 restart associated with it, not TOPLEVEL and ABORT as it used to.
1257 TOP and TOPLEVEL are now available as debugger commands for
1258 returning to the top level.
1259 * incompatible change: forms evaluated in the REPL now use the
1260 global optimization policy.
1261 * incompatible change: user- and system-initialization files are
1262 no longer processed with LOAD, but by READ and EVAL; hence the
1263 global optimization policy, startup package, readtable, etc,
1265 * The .fasl file format number has been incremented because of
1266 various incompatible changes.
1267 * internal entry point details and argument counts no longer appear
1268 in backtraces unless explicitly requested by setting
1269 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1270 * built-in and standard functions no longer have names like "top
1271 level local call to FOO".
1272 * fixed bug 32: functions defined in non-null lexical environments
1273 now have more legible printed representation
1274 * fixed bug 33: functions defined in non-null lexical environemnts
1275 are now more amenable to inspection by INSPECT.
1276 * workaround for bug 354: XEPs no longer appear in backtraces unless
1277 explicitly requested.
1278 * fixed bug: receiving the signal which results from attempting to
1279 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1280 SIGBUS on other *BSDs) on architectures where the C stack is also
1281 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1282 to debugger confusion or wild instability. (reported by Cheuksan
1284 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1285 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1286 (reported by Lutz Euler)
1287 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1288 compiled code. (reported by Paul Dietz)
1289 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1290 specialized parameter to a method to something that is not TYPEP
1291 the specializer is now possible.
1292 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1293 face of package deletion.
1294 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1295 pathname. (reported by David Tolpin on c.l.l)
1296 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1297 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1298 STRING-STREAM and FILE-STREAM, respectively.
1299 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1300 than previously; in addition, its test suite is now run on build.
1301 (thanks to Luke Gorrie)
1302 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1303 * optimization: added a immediate representation for single-floats
1305 * fixed some bugs revealed by Paul Dietz' test suite:
1306 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1307 ** attempting to create a package with a colliding nickname causes
1308 correctable errors to be signalled.
1309 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1310 ** several x86-64 backend bugs related to sign-extension of immediate
1313 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1314 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1315 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1317 * fixed loading of multiply forward-referenced layouts.
1318 (thanks to Cheuksan Wang)
1319 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1320 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1321 (see COMPILE-FILE documentation for details of :PRINT option)
1322 * fixed bugs 19 and 317: fixed-format floating point printing is
1323 more accurate. This also fixes a bug reported by Adam Warner
1324 related to the ~@F format directive.
1325 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1327 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1328 dispatch table if the from-char is a dispatch macro character.
1329 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1330 by the wordsize no longer ignore the last word. (reported by Lutz
1332 * fixed bug in type checking of dynamic-extent variables. (reported
1334 * optimization: sequence traversal functions use their freedom to
1335 coerce function designators to functions.
1336 * optimization: code with many calls to CLOS methods specialized on
1337 CLOS classes has had redundant type checks removed.
1338 * optimization: added declarations to speed up operations that access
1339 the internal character database (for example STRING-UPCASE)
1340 * optimization: comparison operations between floats and sufficiently small
1341 fixnums no longer create extra rationals
1342 * fixed some bugs related to Unicode integration:
1343 ** portions of multibyte characters at the end of buffers for
1344 character-based file input are correctly transferred to the
1345 start of the buffer at the next read.
1346 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1347 passing it through to OPEN.
1348 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1349 argument given to internal calls to OPEN.
1350 ** the built-in comment readers (introduced by character sequences
1351 ";" and "#|") are more forgiving to encoding errors; they will
1352 STYLE-WARN and then attempt to resync the stream at a character
1353 boundary. (thanks to Teemu Kalvas)
1354 * fixed some bugs in the x86-64 port:
1355 ** Negative short int return values from c-calls are sign-extended
1357 ** The stack is aligned to 16-bytes for c-calls, as required by
1358 the ABI. (thanks to Cheuksan Wang)
1359 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1360 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1361 for enums. (thanks to Vincent Arkesteijn)
1362 ** Multiple small optimizations and bugfixes for floating point
1364 * fixed some bugs revealed by Paul Dietz' test suite:
1365 ** Space, Tab, Linefeed, Return and Page have the invalid
1366 secondary constituent character trait.
1367 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1369 ** WITH-INPUT-FROM-STRING should only update the index place on
1371 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1372 works more reliably.
1373 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1374 with one-dimensional arrays with fill pointers.
1375 ** TYPE-OF failed on a complex with an integer realpart and a
1377 ** compiler failure during type inference for the code of form
1378 (IF (EQL X (THE ...)) ...) (MISC.535).
1380 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1381 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1382 platform. The port supports 61-bit fixnums, large memory spaces
1383 and reloading shared object files.
1384 * enhancement: saving cores with foreign code loaded is now
1385 supported on ppc/Darwin in addition to the previously supported
1387 * enhancement: the statistical profiler now walks deeper into the
1388 call stack for more meaningful call-graphs and accrued time
1389 reports (x86/x86-64 only). It also now reports time spent in
1391 * enhancement: it is now possible to trace most individual methods
1392 of a generic function in addition to tracing the generic function
1394 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1396 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1397 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1398 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1399 produces an error. (thanks to Vincent Arkesteijn)
1400 * bug fix: NAMESTRING on pathnames with :WILD components in their
1401 directories works correctly. (thanks to Artem V. Andreev)
1402 * fixed bug 125: compiler preserves identity of closures. (reported
1404 * bug fixed: functions with &REST arguments sometimes failed with
1405 "Undefined function" when compiled with (DEBUG 3). (reported by
1407 * bug fix: overflow during compiling of setting element of a bit
1408 vector with constant index and value. (reported by Timmy Douglas)
1409 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1410 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1411 * on x86 compiler partially supports stack allocation of dynamic-extent
1413 * GO and RETURN-FROM do not check the extent of their exit points
1414 when compiled with SAFETY 0.
1415 * fixed some bugs related to Unicode integration:
1416 ** encoding and decoding errors are now much more robustly
1417 handled; it should now be possible to recover even from invalid
1418 input or output to the terminal. (thanks to Teemu Kalvas)
1419 ** provided a first cut at implementing STRING-TO-OCTETS and
1420 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1421 ** altered the SB-MD5 contributed module slightly, changing the
1422 interface just enough for it to be supportable for builds where
1423 lisp characters are not eight bits.
1424 * fixed some bugs revealed by Paul Dietz' test suite:
1425 ** the FORMATTER-generated functions for ~V[ conditionals require
1426 the correct number of arguments.
1427 ** READ-FROM-STRING returns the mandated second value when applied
1428 to displaced strings.
1429 ** the #\Rubout and #\Backspace characters are treated as invalid
1430 constituent characters by the tokenizer.
1432 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1433 * new feature: reloading changed shared object files with
1434 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1435 * new feature: references to foreign variables and functions
1436 can now be compiled and loaded before the corresponding shared
1437 object file is loaded, as long as the foreign definitions are
1438 available at runtime.
1439 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1440 supported. (thanks to Dan Debertin)
1441 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1442 just keywords. (thanks to Vincent Arkesteijn)
1443 * fixed bug #331: structure-class instances corresponding to
1444 DEFSTRUCT forms are now created eagerly.
1445 * fixed bug #345: backtraces from calls to undefined functions work
1446 on x86 as well. Related bug #61 is now also partially fixed on x86
1447 (backtraces from throws to unknown catch tags.)
1448 * bug fix: lambda-list parsing is now stricter vrt. order and number
1449 of lambda-list keywords.
1450 * bug fix: as specified by AMOP, an error is signalled if a
1451 class-option appears multiple times in a DEFCLASS form. (reported
1453 * bug fix: RANDOM can be compiled when the compiler derives the type
1454 of its numeric argument as a disjoint set of small integers.
1455 (reported by Paul Dietz)
1456 * bug fix: starting a core saved with shared objects loaded when
1457 those objects are not available no longer causes threaded SBCL to
1458 hang. (reported by Sean Ross)
1459 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1461 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1462 argument, which must always be NIL. (reported by Kalle Niemitalo)
1463 * bug fix: printing 1.0d+23 no longer results in an error.
1464 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1465 * bug fix: structure slot setters preserve evaluation order. (thanks
1467 * fixed some bugs related to Unicode integration:
1468 ** RUN-PROGRAM can allow its child to take input from a Lisp
1469 stream. (reported by Stefan Scholl)
1470 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1471 locales. (reported by Ken Causey)
1472 * fixed some bugs revealed by Paul Dietz' test suite:
1473 ** INCF, DECF and REMF evaluate their place form as specified in
1475 ** FORMATTER expands ~{ ~} iteration directives with V or #
1476 parameters correctly.
1477 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1478 consequent uses no arguments correctly.
1479 ** the system has a partial understanding of the (COMPLEX RATIO)
1481 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1484 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1485 * new feature: a build-time option (controlled by the :SB-UNICODE
1486 keyword feature, enabled by default) for building the system with
1487 support for the entire 21-bit character space defined by the
1489 * new feature: the system now has rudimentary external-format
1490 support; the primary user-visible change at this time is that
1491 characters with the high bit set (such as lower-case-e-acute) will
1492 print correctly to a terminal in a UTF-8 environment.
1493 * minor incompatible change: BASE-CHAR no longer names a class;
1494 however, CHARACTER continues to do so, as required by ANSI.
1495 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1496 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1497 has been moved to the SB-EXT package (temporarily re-exported from
1499 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1500 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1502 * fixed bug #351: better error-handlind and reporting for malformed
1504 * fixed bug #350: bignum-printing is now more memory-efficient,
1505 allowing printing of very large bignums, eg. (expt 2 10000000).
1506 (reported by Bruno Haible)
1507 * fixed bug #302: better primitive-type selection for intersection
1509 * fixed bug #308: non-graphic characters now all have names, as
1510 required. (reported by Bruno Haible)
1511 * bug fix: redefining a class with different superclasses now correctly
1512 removes it from the direct-subclasses of its previous superclasses.
1513 (reported by David Morse)
1514 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1515 new value now works. (reported by Bruno Haible)
1516 * bug fix: correct canonicalization of multiple non-standard slot
1517 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1518 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1519 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1520 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1521 now exists, an signals an error.
1522 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1523 during macroexpansion for non-iteration variables as well. (reported
1524 by Bruno Haible for CMUCL)
1525 * bug fix: Cyclic structures and unprintable objects in compiler
1526 messages no longer cause errors. (reported by Bruno Haible)
1527 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1528 and READ-FROM-STRING all now return a primary value of NIL if
1529 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1530 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1531 (reported by Bruno Haible for CMUCL)
1532 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1533 do not designate a condition. (reported by Bruno Haible for
1535 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1536 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1537 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1538 permitting resolution in favour of any of the conflicting symbols.
1539 (reported by Bruno Haible for CMUCL)
1540 * bug fix: EQUAL compiler optimizations is less aggressive on
1541 strings which can potentially compare true despite having distinct
1542 specialized array element types.
1543 * bug fix: unit enumerations can be defined without dividing by
1544 zero. (thanks to Vincent Arkesteijn)
1545 * FORMAT compile-time argument count checking has been enhanced.
1546 (report from Bruno Haible for CMUCL)
1547 * a partial workaround for the bug 262: the compiler does not try to
1548 inline-expand a local function doing RETURN-FROM from a deleted
1549 BLOCK. (thanks to Peter Denno for the bug report and to David
1550 Wragg for the simple test case)
1551 * fixed some bugs revealed by Paul Dietz' test suite:
1552 ** RENAME-PACKAGE allows all package designators as new package
1554 ** constraint propagation and lambda variable substitution are
1555 more cautious in dealing with partially deleted code.
1556 ** compiler sometimes lost reoptimization passes.
1557 ** CERROR, given a condition as condition designator, passes its
1558 remaining arguments to the continue format control without
1560 ** Case-altering FORMAT directives work correctly on non-ASCII
1562 ** The REAL type specifier handles bounds outside the floating
1563 point ranges without signalling FLOATING-POINT-OVERFLOW.
1564 ** Functions with IR1-transformations can create intercomponent
1565 references to global functions.
1566 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1568 ** FORMAT ~R treats a nil value for its first parameter correctly.
1570 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1571 * enhancement: saving cores with foreign code loaded is now
1572 supported on x86/NetBSD and sparc/Linux in addition to the previously
1573 supported platforms.
1574 * bug fix: on some platforms repeated installations caused multiple
1575 copies of HTML documentation to be installed -- should not happen
1576 any more. (reported by Stefan Scholl)
1577 * bug fix: parsing self-recursive alien record types multiple times
1578 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1579 original patch by Helmut Eller for CMUCL)
1580 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1581 (thanks to Richard Kreuter)
1582 * bug fix: defining classes whose accessors are methods on existing
1583 generic functions in other (locked) packages no longer signals
1584 bogus package lock violations. (reported by François-René Rideau)
1585 * bug fix: special variables as DEFMETHOD parameters no longer have
1586 associated bogus type declarations. (reported by David Wragg and
1588 * bug fix: read-write consistency on streams of element-type
1589 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1590 * bug fix: redefiniton of the only method of a generic function with
1591 no DEFGENERIC no longer emits a full WARNING. In addition,
1592 redefinition of generic functions with no DEFGENERIC to an
1593 incompatible lambda list now signals an error. (thanks to Zach
1595 * bug fix: DEFGENERIC now works even when there's a function of the
1596 same name in an enclosing lexical environment. (thanks to Zach
1598 * fixed compiler failure, caused by instrumenting code during
1599 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1600 * optimization: added loop analysis and improved register allocation
1601 * fixed some bugs revealed by Paul Dietz' test suite:
1602 ** POSITION on displaced vectors with non-zero displacement
1603 returns the right answer.
1604 ** (SIMPLE-STRING) is a valid type specifier for sequence
1606 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1608 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1609 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1611 ** The justification version of the FORMAT ~< directive treats
1612 non-zero minpad parameter correctly.
1614 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1615 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1616 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1617 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1618 the supported interface.
1619 * new feature: Single-stepping of code compiled with DEBUG 2 or
1620 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1621 * new feature: saving cores with foreign code loaded is now
1622 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1623 Timothy Moore's work for CMUCL)
1624 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1625 parameters to * instead of NIL if no initform is supplied.
1626 (reported by Johan Bockgaard)
1627 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1628 correctly when FUNCALL appears as the car of the form. Note:
1629 despite this FUNCALL forms are not currently subject to
1630 compiler-macro expansion. (port of Raymond Toy's fix for the
1631 same from CMUCL, reported by Johan Bockgaard)
1632 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1633 (thanks for Teemu Kalvas)
1634 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1635 case instead of terminating the iteration (thanks for Julian
1636 Squires, Sean Champ and Raymond Toy)
1637 * bug fix: incorrect expansion of defgeneric that caused a style
1638 warning. (thanks for Zach Beane)
1639 * bug fix: slot accessor effective method computation works properly
1640 for classes with multiple non-standard applicable methods on
1641 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1642 * on x86 compiler supports stack allocation of results of LIST and
1643 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1644 CMUCL implementation by Gerd Moellmann)
1645 * fixed some bugs revealed by Paul Dietz' test suite:
1646 ** FORMAT strings with both the ~<~:;~> form of the justification
1647 directive and pretty-printing directives cause an error.
1649 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1650 * incompatible change: the internal functions
1651 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1652 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1653 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1654 instead of the old functions.
1655 * new feature: on platforms where "dladdr" is available foreign
1656 function names now appear in backtraces. (based on Helmut Eller's
1658 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1659 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1661 * documentation: ASDF manual is now distributed with SBCL.
1662 * bug fix: SBCL can now load its contributed modules with REQUIRE
1663 even if the system-provided entries have been removed from
1664 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1666 * bug fix: backtraces involving undefined functions or assembly
1667 routines are more informative. (thanks to Brian Downing)
1668 * bug fix: mutually referent alien structures now work correctly.
1669 (reported by Rick Taube)
1670 * bug fix: structures defined by WITH-ALIEN can be referred to
1671 within other definitions in the same WITH-ALIEN.
1672 * bug fix: division operators (MOD, TRUNCATE and the like) with
1673 constant zero divisors and integer dividends no longer generate
1675 * bug fix: provide default methods for INPUT-STREAM-P and
1676 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1677 * bug fix: improve the detection and resolution of MOP metacycles
1678 (where effective method computation depends on the generic function
1679 having its effective method computed). (reported by Bruno Haible)
1680 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1681 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1682 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1683 * bug fix: pathnames with relative directory components can be
1684 represented relative to default pathnames.
1685 * optimization: in taking the GCD of bignums, reduce the two bignums
1686 to approximately the same size (using Euclid's algorithm) before
1687 applying the more sophisticated binary GCD. (thanks to Juho
1689 * optimization: COUNT on bitvectors now operates word-at-a-time.
1690 * optimization: ASH with a positive, but not necessarily constant,
1691 (leftwards) shift, when in a modular context, is compiled to a
1693 * fixed some bugs revealed by Paul Dietz' test suite:
1694 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1695 if the corresponding argument is NIL.
1696 ** FORMAT directives accepting a minpad parameter treat negative
1698 ** literal commas as character parameters to FORMAT directives are
1700 ** literal spaces directly after ~<Newline> directives within a
1701 format-logical-block (~:< ~@:>) do not induce :FILL-style
1702 conditional newlines.
1703 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1704 as specified: it no longer includes conditional newlines).
1705 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1707 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1708 :PER-LINE-PREFIX argument does not evaluate to a string.
1710 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1711 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1712 the manual for details; for now, package locks can be disabled by
1713 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1714 if no major problems are found then it is likely that they will be
1715 compiled in unconditionally.
1716 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1717 unsupported operators on all platforms. To load a shared library
1718 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1719 object file, link it into a shared library outside of SBCL and
1720 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1722 * bug workaround: ROOM T and the GC were not working together
1723 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1724 reporting facility aren't preserved. That reporting has been
1725 disabled, so now until and unless someone figures out how to make it
1726 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1727 * minor incompatible change: as threatened around sbcl-0.8.0, the
1728 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1729 The symbols are also exported from SB-PCL for backwards
1730 compatibility, but more so than before SB-PCL should be treated as
1731 an implementation-internal package.
1732 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1734 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1735 It is known as of this release not to work on the Alpha, however.
1736 * fixed bug #167: errors signalled due to illegal syntax in method
1737 bodies are now more legible.
1738 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1739 designators and can hence be used with TYPEP.
1740 * fixed bug #333: CHECK-TYPE now ensures that the type error
1741 signalled, if any, has the right object to be accessed by
1742 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1743 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1744 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1746 * fixed bug #334: programmatic addition of slots using specialized
1747 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1748 and :ALLOCATION :CLASS slots.
1749 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1750 float exponents. (rereported by Peter Seibel)
1751 * fixed a bug: #\Space (and other whitespace characters) are no
1752 longer considered to be macro characters in standard syntax by
1753 GET-MACRO-CHARACTER.
1754 * fixed bug: initialization of condition class metaobjects no longer
1755 causes an instance of the condition to be created. (reported by
1757 * fixed bug: it is now possible to have slots such that
1758 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1759 description is neither :INSTANCE nor :CLASS.
1760 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1761 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1762 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1763 non-standard SLOT-DEFINITION-ALLOCATION values.
1764 * fixed bugs: various race conditions handling exiting threads.
1765 CL-PPCRE's thread torture test now passes.
1766 * fixed arguably-a-bug: GC time proportional to number of threads in
1767 system even when most of them are idle
1768 * optimization: improved performance of BIT and SBIT on bit-vectors.
1769 * .fasl file incompatibility: The fasl file version number has
1770 been incremented because of changes associated with package locks.
1772 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1773 * minor incompatible change: the system no longer provides
1774 optimization hints (conditions which are TYPEP
1775 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1776 mode; these hints are still emitted when higher SPEED optimization
1778 * new contrib module: a sampling profiler (profiling by statistical
1779 sampling, rather than by instrumenting functions) is available as
1780 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1781 * the behaviour of the standard function ED is now customizeable by
1782 third parties through a hook variable: see ED's documentation
1783 string for information on the protocol.
1784 * the compiler no longer emits efficiency notes for (FUNCALL X)
1785 when the type of X is uncertain under default optimization
1787 * fixed bug 276: mutating a binding of a specialized parameter to a
1788 method to something that is not TYPEP the specializer is now
1790 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1791 exhibit the required behaviour on the x86 platform. (thanks to
1792 Peter van Eynde, Eric Marsden and Bruno Haible)
1793 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1794 even for difficult arguments. (reported by Peter Graves)
1795 * fixed bug 141a: the backquote printer now descends quoted
1797 * fixed another bug in backquote printing: no more destructive
1798 modification of the form's list structure. (reported by Brian
1800 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1801 move between its address being taken and the call to
1802 interrupt_thread, fixing a crashing race condition.
1803 * the SB-POSIX contrib implementation has been adjusted so that it
1804 no longer exhibits ridiculously poor performance when constructing
1805 instances corresponding to C structs.
1807 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1808 * minor incompatible change: the sb-grovel contrib now treats C
1809 structures as alien (in the sense of SB-ALIEN) objects rather than
1810 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1811 has implications for memory management of client code
1812 (sb-grovel-returned objects must now be manually managed) and for
1813 type safety (alien objects now have full types).
1814 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1815 used to control emission of compiler diagnostics, rather than the
1816 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1817 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1818 quality should be considered deprecated.
1819 * install.sh now installs the user manual as well
1820 * (not quite a new documentable feature, but worth considering in
1821 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1822 beginnings of a semantically meaningful condition hierarchy is
1823 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1825 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1826 designator as the defaults argument.
1827 * fixed bug: Displaced arrays whose displaced-to array has become
1828 too small now cause ARRAY-DIMENSION to signal an error, providing
1829 for safer bounds-checking. (reported by Bruno Haible)
1830 * fixed bug: DEFCLASS slot definitions with identical :READER and
1831 :WRITER names now signal a reasonable error. (reported by Thomas
1833 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1835 * fixed bug: Invalid slot specification errors now print correctly.
1836 (thanks to Zach Beane)
1837 * fixed bug 320: Shared to local slot value transfers in class
1838 redefinitions now happen corrently from superclasses as
1839 well. (reported by Bruno Haible)
1840 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1842 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1843 improper lists correctly. (reported by Bruno Haible)
1844 * fixed bug 313: source-transform for <fun-name> was erroneously
1845 applied to a call of a value of a variable with name <fun-name>.
1846 (reported by Antonio Menezes Leitao)
1847 * fixed bug 307: The obsolete instance protocol ensures that
1848 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1849 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1850 remove dead unknown-values globs from the middle of the stack.
1851 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1852 easier to resume long computations after using *BREAK-ON-SIGNALS*
1853 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1854 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1855 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1856 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1857 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1858 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1860 * fixed a (fairly theoretical) bug in string printing: if
1861 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1862 does not have array-element-type equal to the most general string
1864 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1865 function name. (thanks to Nikodemus Siivola)
1866 * fixed bug: compile-time format string checker failed on
1867 non-closed ~{. (reported by Thomas F Burdick)
1868 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1869 don't warn on legal loop code involving more than one aggregate
1870 boolean. (thanks to Nikodemus Siivola)
1871 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1872 in keyword position to MAKE-PACKAGE was not being reported as
1874 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1875 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1877 * fixed bugs 280 and 312: the checking for multiple definitions in a
1878 file is less likely to become confused by uses of inline
1880 * fixed bug: the #S reader macro performs the keyword coercion
1881 specified for slot names. (reported by Kalle Niemitalo)
1882 * fixed bug: lambda lists may contain symbols whose names start with
1883 & but are not lambda-list-keywords; their occurrence triggers a
1885 * fixed bug 321: define-method-combination argument lambda lists do
1886 not cause invalid code to be generated when &optional or &aux
1887 variables have default values. (reported by Bruno Haible)
1888 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1889 have CLOS classes; this appears also to have fixed reported
1890 instability in stack exhaustion detection.
1891 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1892 MUFFLE-WARNING when no associated restart is present is now
1894 * optimization: rearranged the expansion of various defining macros
1895 so that each expands into only one top-level form in a
1896 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1898 * optimization: used a previously little-used slot in symbols to
1899 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1901 * fixed some bugs revealed by Paul Dietz' test suite:
1902 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1904 ** Fixed an optimization bug related to inheritance of initforms
1905 from local to shared slots.
1906 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1907 ** CAST optimizer forgot to flush argument derived type.
1908 ** print/read consistency on floats is now orders of magnitude
1909 more likely. (thanks also to Bruno Haible for a similar report
1911 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1912 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1913 for objects of type REAL. Make it so.
1914 ** SXHASH is specified (infelicitously) to respect similarity,
1915 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1916 it so. (thanks to Markus Ziegler)
1917 ** on the Alpha, the compiler succeeds in compiling functions
1918 returning a known number of arguments greater than 63.
1919 ** fixed handling of invalid NIL arguments in keyword position
1921 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1923 ** provided a readably-printable representation for RANDOM-STATE
1925 ** ensured that pathnames, where they have a namestring, always
1926 print using #P"..." syntax.
1928 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1929 * Support for the forthcoming 2.0 version of the NetBSD kernel
1930 running on x86 hardware has been added. (thanks to Perry
1931 E. Metzger most immediately, and others for their past work)
1932 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1933 current release version; SBCL's OpenBSD support had been broken
1934 since about the time of OpenBSD's switch to ELF binary format.)
1935 * [placeholder for DX summary]
1936 ** user code with &REST lists declared dynamic-extent, under high
1937 speed or space and low safety and debug optimization policy.
1938 * The manual has been converted to Texinfo format and the debugger
1939 chapter from the cmucl manual has been added.
1940 * A facility has been added to extract documentation strings from
1941 sbcl and store them as Texinfo-formatted snippets for inclusion in
1942 the manual (via Texinfo's @include directive)
1943 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1944 the test case to Dave Roberts)
1945 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1946 pointers. (reported by Sean Ross)
1947 * bug fix: PROFILE output is printed nicely even for large numerical
1948 values. (thanks to Zach Beane)
1949 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1950 greater than 32 handle EOF correctly.
1951 * bug fix: on X86 an immediate argument of the IMUL instruction is
1952 correctly printed in disassembly. (thanks to Lutz Euler)
1953 * bug fix: class slots in redefined classes preserve their old
1954 values. (thanks to Bruno Haible and Nikodemus Siivola)
1955 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1956 succeeds. (reported by Marco Baringer)
1957 * fixed some bugs revealed by Paul Dietz' test suite:
1958 ** READ-SEQUENCE now works on ECHO-STREAMs.
1959 ** RATIONALIZE works more according to its specification. (thanks
1962 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1963 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1964 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1965 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1966 should matter to you only if you rebind the printer control
1967 variables and then find you want different bindings in the
1968 debugger than in the ordinary execution of your program.)
1969 * The runtime build system has been tweaked to support building
1970 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1971 assembler and linker. (thanks to Nikodemus Siivola)
1972 * Unbound, undefined, undeclared variables now trigger full
1973 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1974 more often programmer error than deliberate exploitation of undefined
1976 * optimization: the hash algorithm for strings has changed to one
1977 that is less vulnerable to spurious collisions. (thanks to Juho
1979 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1980 needless bounds checking. (thanks to Juho Snellman)
1981 * optimization: implemented multiplication as a modular
1982 (UNSIGNED-BYTE 32) operation on the PPC backend.
1983 * fixed some bugs revealed by Paul Dietz' test suite:
1984 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1986 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1987 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1988 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1990 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1992 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1994 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1995 their output stream on EOF from read.
1996 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1997 have been read to end-of-file.
1998 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2000 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2001 description of determination of which consecutive characters
2003 ** Printing the "Space" character with escaping on now yields "#\\ ",
2004 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2005 ** Reading floating-point numbers with *READ-BASE* set to a number
2006 less than 10 works correctly.
2007 ** Reading floating-point numbers with *READ-BASE* set to a number
2008 more than 10 works correctly.
2009 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2010 the readtable currently in effect.
2012 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2013 * minor incompatible change: parsing of namestrings on a physical
2014 (Unix) host has changed; numbers after the final #\. in a
2015 namestring are no longer interpreted as a version field. This is
2016 intented to be largely invisible to the user, except that the
2017 meaning of the namestring "*.*.*" has changed: it now refers to a
2018 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2019 should usually be replaced by
2020 (make-pathname :name :wild :type :wild :version :wild)
2021 with the added benefit that this is more likely to be portable.
2022 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2023 option to OPEN now signals an error if the file being opened
2024 exists; this may have an impact on existing code.
2025 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2026 signals received, on the PowerPC platforms (both Linux and
2027 Darwin). (thanks to Pierre Mai for pointing out the location of
2029 * several fixes on OS X: The system now builds and runs cleanly on
2030 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2031 (thanks to Brian Mastenbrook)
2032 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2033 second-resolution: integer multiples of 1/3600 between -24 and 24.
2034 (thanks to Vincent Arkesteijn)
2035 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2036 type when called with 1 argument; PEEK-CHAR checked type of
2037 PEEK-TYPE only after having read first character from a
2038 stream. (reported by Peter Graves)
2039 * bug fix: the garbage collector now has much better locality
2040 behaviour, and in particular no longer treats all memory as being
2041 exhausted when many small objects point to each other in a deeply
2043 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2044 recognized as being TYPEP their class.
2045 * bug fix: the PUSHNEW documentation string has been corrected.
2046 (thanks to Vincent Arkesteijn)
2047 * bug fix: defaulting of the value for the last of an atypically
2048 large number of multiple values being bound was not being
2049 performed correctly on the Alpha or PPC platforms
2050 * optimization: implemented multiplication as a modular
2051 (UNSIGNED-BYTE 32) operation on the x86 backend.
2052 * optimization: SEARCH on simple-base-strings can now be open-coded.
2053 (see also contrib/compiler-extras.lisp for inspiration for
2054 teaching the compiler about the Boyer-Moore algorithm).
2055 * value, returned by MAX (and MIN) called with several EQUALP, but
2056 not EQL, arguments now does not depend on compiler settings.
2057 * fixed some bugs revealed by Paul Dietz' test suite:
2058 ** in stack analysis liveness information is propagated from
2059 non-local entry points.
2060 ** pathwise CAST removing failed when the CAST node did not start
2062 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2063 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2065 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2066 host is already defined.
2067 ** RENAME-FILE works on streams instead of signalling an internal
2069 ** PEEK-CHAR uses the current readtable when determining whether
2070 or not a character is whitespace.
2071 ** MERGE-PATHNAMES handles the case when the pathname does not
2072 specify a name while the default-pathname specifies a version.
2073 ** Pathnames now stand a better chance of respecting print/read
2075 ** Attempting to use standardized file system operators with a
2076 pathname with invalid :DIRECTORY components signals a
2078 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2080 ** DIRECTORY on logical pathnames is more correct.
2081 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2082 signal a TYPE-ERROR if their argument is not a stream.
2083 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2084 designator argument does not designate a stream.
2085 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2086 examining the synonym.
2087 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2089 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2090 element-type larger than ([UN]SIGNED-BYTE 32).
2092 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2093 * When built with the :SB-FUTEX feature, threaded builds now take
2094 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2095 for faster/more reliable mutex and condition variable support.
2096 * Incompatible change (but one you probably shouldn't have been using
2097 anyway): the interface and code for arbitrating between multiple
2098 threads in the same user session has been redesigned.
2099 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2100 error when called without an explicit environment argument.
2101 (thanks to Nikodemus Siivola)
2102 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2103 confused about their position. (thanks to Adam Warner and Gerd
2105 * bug fix: Pretty printing backquoted forms with unquotations in the
2106 argument list position of various code constructs such as LAMBDA
2107 now works correctly. (reported by Paul Dietz)
2108 * bug fix: Pretty printing unquotations no longer loses all
2109 stream position information.
2110 * optimization: performance of string output streams is now less
2111 poor for multiple small sequence writes.
2112 * optimization: performance of CSUBTYPEP in the presence of complex
2113 expressions involving CONS and NOT many times has been improved.
2114 (reported by Paul Dietz)
2115 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2117 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2118 streams for the REPL, for the debugger, and so on. (thanks to
2120 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2121 * fixed some bugs revealed by Paul Dietz' test suite:
2122 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2123 is ignored if it is too small, rather than propagating through
2124 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2125 ** extremely complex negations of CONS types were not being
2126 sufficiently canonicalized, leading to inconsistencies in
2128 ** VALUES tranformer lost derived type.
2130 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2131 * fixed a bootstrapping bug: the build process no longer assumes
2132 that the various BOOLE-related constants have the same value in
2133 host and target lisps. (noted by Paul Dietz' test suite on an
2134 SBCL binary built from CLISP)
2135 * The system can now be dynamically linked on the MIPS platform,
2136 which enables dynamic loading of foreign code from Lisp. (thanks
2137 to Ralf Baechle for discussions on the MIPS ABI)
2138 * The system now records debugging information for its own source
2139 files in a filesystem-position-independent manner, relative to
2140 the "SYS" logical host.
2141 * fixed a compiler bug: MV-LET convertion did not check references
2142 to the "max args" entry point. (reported by Brian Downing)
2143 * tweaked disassembly notes to be less confident about proclaiming
2144 some instruction as an LRA. (thanks to Brian Downing)
2145 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2146 now each have their own history, command character, and other
2147 characteristics. (thanks to David Lichteblau)
2148 * fixed some bugs revealed by Paul Dietz' test suite:
2149 ** compiler failure in compiling LOGAND expressions including a
2151 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2152 shift greater than 32.
2153 ** FLUSH-DEST did not mark blocks for type check regeneration.
2154 ** HANDLER-CASE failed to accept declarations in handler clauses
2155 in some circumstances.
2157 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2158 * New code in contrib/sb-introspect (still probably not entirely
2159 stable yet) provides some support for smart Lisp development
2160 environments like SLIME.
2161 * The conditions signalled for errors occurring when loading .fasl
2162 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2163 in a way which should help ASDF recover gracefully.
2164 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2165 stuff has been cleaned up. If you code contrib/ stuff, this might
2166 affect you, and you can look at contrib/README, contrib/STANDARDS,
2167 and/or the 0.8.4.27 diff to check.
2168 * In full calls the compiler now does not generate checks for declared
2169 argument types for all arguments.
2170 * various threading fixes
2171 ** and some experimental patches which didn't make it into
2172 the main tree for this release, but which are shipped in
2173 contrib/experimental-thread.patch as a possible fix for some
2174 failures (deadlock, spinning...) in GC-intensive multithreaded
2176 * fixed PPC build problem (source code incompatibility of different
2177 library versions): added offsetof() hackery which attempts to divine
2178 where glibc maintainers put uc_mcontext today
2179 * fixed bug 282: compiler does not trust type assertions while passing
2180 arguments to a full call.
2181 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2182 &OPTIONAL VALUES type parameter.
2183 * fix bug 214: algorithm for noting rejected templates is now more
2184 similar to that of template seletion. (also reported by rydis on
2186 * fixed bug 141b: printing backquoted information readably and prettily
2187 inserts a space where necessary.
2188 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2189 (CHARACTER 10) are now reported as errors, rather than propagated
2190 as unknown types. (reported by piso on #lisp)
2191 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2192 with values NIL and :ERROR. (thanks to Milan Zamazal)
2193 * fixed bug 191c: CLOS now does proper keyword argument checking as
2194 described in CLHS 7.6.5 and 7.6.5.1.
2195 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2196 longer raise an error; further, using a list as a for-as-arithmetic
2197 counter now raises a meaningful error.
2198 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2199 now understood by sequence creation functions such as MAKE-SEQUENCE
2201 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2202 a STRING-INPUT-STREAM.
2203 * compiler enhancement: SIGNUM is now better able to derive the type
2205 * type declarations inside WITH-SLOTS are checked. (reported by
2207 * fixed some bugs revealed by Paul Dietz' test suite:
2208 ** incorrect optimization of TRUNCATE for a positive first
2209 argument and negative second.
2210 ** compiler failure in let-convertion during flushing dead code.
2211 ** compiler failure while deriving type of TRUNCATE on an
2212 interval, containing 0.
2213 ** ASH of a negative bignum by a negative bignum count now returns
2215 ** intersection of CONS types now canonicalizes properly, fixing
2216 inconsistencies in SUBTYPEP.
2218 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2219 * incompatible change: The --disable-debugger command line
2220 option now clobbers the debugger at a more fundamental
2221 level, by redefining #'INVOKE-DEBUGGER instead of by
2222 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2223 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2224 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2225 enter the debugger REPL and then suffer recursive errors
2226 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2227 while under the new system, BREAK in --disable-debugger mode
2228 terminates the system just as an unhandled error would.
2229 * fixed compiler performance when processing loops with a step >1;
2230 * bug fix: DOCUMENTATION now retrieves generic function
2231 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2232 support has been systematized, and now supports the methods
2233 specified by ANSI, along with a default method and a method for
2234 slot documentation. (reported by Nathan Froyd)
2235 * bug fix: effective methods associated with a generic function are
2236 no longer cached over a change of that generic function's method
2237 combination. (reported by Andreas Fuchs)
2238 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2239 names a non-existent directory. (thanks to Andreas Fuchs)
2240 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2241 the PPC platform, lead to stack corruption; this has been fixed.
2242 (reported by Rainer Joswig)
2243 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2244 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2245 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2246 arithmetic in general, is now much more reliable on the PPC
2248 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2249 platform now returns the right answer.
2250 * optimization: restored some effective method precomputation in
2251 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2252 precomputation is now tunable.
2253 * optimization: compiler-internal data structure use has been
2254 reviewed, and changes have been made that should improve the
2255 performance of the compiler by about 20%.
2256 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2257 simple-base-strings and simple-bit-vectors is improved.
2258 * optimization: the optimization of 32-bit logical and arithmetic
2259 functions introduced in version 0.8.3 on the x86 has been
2260 implemented on the mips, ppc and sparc platforms; an
2261 implementation of the same facility, but for 64-bit arithmetic,
2262 has been added for the alpha.
2263 * microoptimization: the compiler is better able to make use of the
2264 x86 LEA instruction for multiplication by constants.
2265 * bug fix: in some situations compiler did not report usage of
2266 generic arithmetic in (SPEED 3) policy.
2267 * bug 145b fix: compiler used wrong type specifier while converting
2268 MEMBER-types to numeric.
2269 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2271 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2272 index argument now works properly on the Alpha platform.
2273 * bug fix: floating point exception treatment on the Alpha platform
2275 * bug fix: FILE-POSITION works much better on string input and
2276 output streams. (thanks to Nikodemus Siivola)
2277 * bug fix: many threading/garbage collection symptoms sorted.
2278 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2279 might be pseudo-atomic.
2280 * internal change: Stopping for GC is now done with signals not
2281 ptrace. GC is now done in whichever thread wanted it, instead of
2283 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2285 * bug fix: result form in DO is not contained in the implicit
2287 * incompatible change: ICR structure is changed; the value part of
2288 CONTINUATION is now called LVAR; corresponding functions are
2289 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2290 * added type deriver for ISQRT (thanks to Robert E. Brown).
2291 * bug fix: better support for loading from the command line when an
2292 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2294 * fixed some bugs revealed by Paul Dietz' test suite:
2295 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2296 ** ROUND and FROUND now give the right answer when given very
2297 small float arguments.
2298 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2300 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2301 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2302 ** LCM should return a non-negative integer.
2303 ** PARSE-INTEGER returned the index of a terminator instead of the
2304 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2305 ** PARSE-INTEGER returned an incorrect index being applied to a
2307 ** LCM with two arguments of 0 returns 0 rather than signalling
2309 ** unsigned addition of a 32-bit constant with the high bit set no
2310 longer causes an internal compiler error on the x86.
2311 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2312 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2313 with negative last argument.
2314 ** byte specifiers with zero size and position no longer cause
2315 an error during type derivation.
2316 ** bignum multiplication on the Alpha platform now returns the
2318 * porting: The system now builds on SuSE AMD64, although it still
2319 generates a 32-bit binary.
2320 * .fasl file incompatibility: The fasl file version number has
2321 been incremented (because of the changes to internal compiler
2322 data structures referred to above).
2324 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2325 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2326 more accurately, on the Darwin kernel running on PowerPC hardware.
2327 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2328 * Compiler code deletion notes now signal a condition of type
2329 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2330 an associated MUFFLE-WARNING restart.
2331 * The compiler now performs limited argument count validation of
2332 constant format strings in FORMAT, and where appropriate in ERROR,
2333 CERROR and WARN. (thanks to Gerd Moellmann)
2334 * New ASDF-INSTALL contrib can be used for automatic download and
2335 installation of third-party Lisp code from CCLAN or other sites
2337 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2338 which forces another thread to execute a function supplied by the
2340 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2341 now accept and act upon their :ELEMENT-TYPE keyword argument.
2342 (reported by Martin Atzmueller, Edi Weitz)
2343 * bug fix: FILE-POSITION now accepts position designators up to
2344 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2345 is the greater. (thanks to Patrik Nordebo)
2346 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2347 Kalle Olavi Niemitalo)
2348 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2349 printable. (reported by Eric Marsden)
2350 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2351 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2352 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2353 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2354 (thanks to Stig Sandoe)
2355 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2356 UNREAD-CHAR now correctly includes the unread character in the
2357 target sequence. (thanks to Gerd Moellmann)
2358 * bug fix in threaded builds: the system can now be suspended and
2359 resumed by shell job control with minimal disruption.
2360 * bug fixes in times and timezones >2038AD
2361 * better handling of "where is GNU make?" problem in build scripts
2362 (thanks to Nikodemus Siivola)
2363 * new optimization: inside a named function any reference to a
2364 function with the same name is considered to be a self-reference;
2365 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2366 optimization quality.
2367 * new optimization on x86: logical functions and + now have
2368 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2369 used when the result is truncated to 32 bits.
2370 * VALUES declaration is partially enabled.
2371 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2372 * fixed some bugs revealed by Paul Dietz' test suite:
2373 ** The system now obeys the constraint imposed by
2374 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2375 types form a lattice under type intersection.
2376 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2377 ** ASSOC now ignores NIL elements in an alist.
2378 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2379 and (1+ MOST-POSITIVE-FIXNUM) answers.
2380 ** The addition of a method with invalid qualifiers to a generic
2381 function does not cause an error to be signalled immediately;
2382 a warning is signalled, and the error is generated only on
2383 calling the generic function.
2384 * changed .fasl file version number, in part to add type codes for
2385 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2386 obscure ANSI requirements
2388 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2389 * fixed bug 148: failure to inline-expand a local function left
2390 garbage, confusing the compiler.
2391 * fixed bugs 3cd: structure slot readers perform type check if the
2392 slot can have an invalid value (i.e. it is either not initialized
2393 or can be written with a less specific slot writer).
2394 * bug fix: the compiler now traps array references to elements off
2395 the end of an array; previously, the bounds checking in some
2396 circumstances could go off-by-one.
2397 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2399 * type declarations for array element types now obey the description
2400 on the CLHS page "Declaration TYPE", as per discussions on
2401 sbcl-help around 2003-05-08. This means that a declaration
2402 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2403 declaration, all references to BAR will be asserted or assumed
2404 (with THE, so dependent on compiler policy) to involve objects of
2405 type FOO. Note that no such declaration is implied in
2406 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2407 * declared types of functions from the "Conditions"
2408 chapter. (reported by Paul Dietz)
2409 * bug fix: CERROR accepts a function as its first argument.
2410 * bug fix: NTH an NTHCDR accept a bignum as index
2411 arguments. (reported by Adam Warner)
2412 * optimization: character compare routines now optimize comparing
2413 against a constant character. (reported by Gilbert Baumann)
2414 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2415 argument now works properly on the MIPS platform.
2416 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2417 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2418 declared to have a functional type, does not cause an error
2420 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2421 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2422 * fixed bug: if last continuation of a deleted block has a
2423 destination, this destination should be deleted too. (reported by
2425 * fixed a bug in the bootstrap process: the host compiler's values
2426 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2427 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2428 test case from Patrik Nordebo)
2429 * improved the ability of the disassembler on the PPC platform to
2430 provide helpful disassembly notes.
2431 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2432 the class in more cases than previously.
2433 * bug fix: FILE-POSITION now understands :START and :END for
2434 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2435 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2436 8) data. (thanks to David Lichteblau)
2437 * bug fix: it is possible to add a method to a generic function
2438 without lambda list.
2439 * bug fix: reader failed to signal END-OF-FILE inside an
2440 object representation. (reported by Nikodemus Siivola)
2441 * fixed some bugs revealed by Paul Dietz' test suite:
2442 ** LAST and [N]BUTLAST should accept a bignum.
2443 ** condition slot accessors are methods.
2444 ** (VECTOR NIL) is a subtype of STRING.
2446 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2447 * minor incompatible change: some nonsensical specialized lambda
2448 lists (used in DEFMETHOD) which were previously ignored now signal
2450 * minor incompatible change: the system is now aware of the types of
2451 variables in the COMMON-LISP package, and will signal errors for
2452 most violations of these type constraints (where previously they
2453 were silently accepted).
2454 * minor incompatible change: COMPILE-FILE now uses the freedom
2455 afforded (ANSI 3.2.2.3) to use derived function types for
2456 functions defined in the same file. This also permits the system
2457 to warn on static type mismatches and function
2458 redefinition. (Currently it does not work with high DEBUG level.)
2459 * minor incompatible change: VALUES declaration is disabled.
2460 * When issuing notes, the compiler now signals a condition of type
2461 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2462 restart for use in user handlers. It is expected that the
2463 COMPILER-NOTE condition will eventually become a condition
2464 supertype to a hierarchy of note types, which will then be
2465 handleable in a similar fashion. However, at the moment, no such
2466 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2467 but it's an implementation detail, not a classification for the
2469 * Changes in type checking closed the following bugs:
2470 ** type checking of unused values (192b, 194d, 203);
2471 ** template selection based on unsafe type assertions (192c, 236);
2472 ** type checking in branches (194bc).
2473 * A short form of VALUES type specifier has ANSI meaning (it has
2474 increased the number of situations when SBCL cannot perform type
2476 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2477 keywords or constants is permissible.
2478 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2479 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2480 classes. (thanks to Antonio Martinez)
2481 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2482 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2483 argument to a pretty-print stream. Instead, it leaves any such
2484 operations to DESCRIBE-OBJECT methods.
2485 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2487 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2488 lambda lists. (thanks to David Lichteblau)
2489 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2490 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2491 (thanks to Gerd Moellmann)
2492 * fixed bug 246: increased compilation speed of long
2493 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2495 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2497 * DEFCONSTANT now throws a condition of type
2498 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2499 constant to a non-EQL value; CONTINUE and ABORT restarts
2500 respectively change and preserve the value.
2501 * fixed bug 63: The code walker, part of the implementation of CLOS,
2502 is now better at handling symbol macros.
2503 * bug fix: There is no longer an internal implementation type named
2504 CL:LENGTH. (reported by Raymond Toy)
2505 * bug fix: In macro-like defining macros/special operators the
2506 implicit block does not enclose the lambda list.
2507 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2508 enclosed in parentheses) are not suitable as type specifiers, and
2509 their use properly signals an error now.
2510 * bug fix: An argument count mismatch for a type specifier in code
2511 being compiled no longer causes an unhandled error at compile
2512 time, but signals a compile-time warning.
2513 * fixed simple vector readable printing
2514 * bug fix: DESCRIBE takes more care over whether the class
2515 precedence list slot of a class is bound before accessing it.
2516 (reported by Markus Krummenacker)
2517 * bug fix: FORMATTER can successfully compile pretty-printer format
2518 strings which use variants of the ~* directive inside.
2519 * bug fix: SEARCH now applies its TEST predicate to the elements of
2520 the arguments in the correct order. (thanks to Wolfhard Buss)
2521 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2522 in type inference. (thanks to Robert E. Brown)
2523 * bug fix: Reading in symbols with an explicit package name of ""
2524 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2525 * fixed some bugs revealed by Paul Dietz' test suite:
2526 ** NIL is now allowed as a structure slot name.
2527 ** Arbitrary numbers, not just REALs, are allowed in certain
2528 circumstances in LOOP for-as-arithmetic clauses.
2529 ** Multiple class redefinitions before slot access no longer
2530 causes a type error.
2531 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2532 association between the name and a class.
2533 ** Generic functions with non-standard method-combination and over
2534 five methods all of which return constants no longer return NIL
2535 after the first few invocations. (thanks to Gerd Moellmann)
2536 ** CALL-NEXT-METHOD with no arguments now passes the original
2537 values of the arguments, even in the presence of assignment.
2538 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2540 ** DISASSEMBLE works with closures and funcallable instances.
2541 ** ADD-METHOD now returns the generic function, not the new method.
2542 ** FIND-METHOD signals an error if the lengths of the specializers
2543 is incompatible with the generic function, even if the ERRORP
2545 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2546 which its argument is a member.
2547 ** DEFCLASS only redefines the class named by its class-name
2548 argument if that name is the proper name of the class;
2549 otherwise, it creates a new class.
2550 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2551 of the UNBOUND-SLOT condition to the name of the slot.
2552 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2553 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2554 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2555 SLOT-MAKUNBOUND in the specified fashion.
2557 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2558 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2559 cross-compilation host. As a consequence, we can now bootstrap our
2560 way up to SBCL starting with a bare gcc toolchain and human-readable
2561 source code (first the source to CLISP, then the source to SBCL).
2562 * A contributed module containing a partial implementation of the
2563 simple-streams interface has been included. (thanks to Rudi
2565 * A contributed module implementing the RFC1321 Message Digest
2566 Algorithm, known as MD5, has been included.
2567 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2568 no longer has any effect, as the code controlled by this feature
2569 has been deleted. (As far as we know, no-one has ever built using
2570 this feature, and its semantics were confused in any case).
2571 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2572 work on conditions (as required by the ANSI specification),
2573 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2574 expected behaviour on conditions. Users should note, however,
2575 that such behaviour is not required by the ANSI specification,
2576 and so use of this behaviour may render their code unportable.
2577 * fixed some bugs revealed by Paul Dietz' test suite:
2578 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2580 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2581 ** SXHASH on condition objects no longer returns NIL.
2582 ** :ALLOCATION :CLASS slots are better treated; their values are
2583 updated on class redefinition, and initforms inherited from
2584 superclasses are applied.
2585 ** REMOVE-METHOD returns its generic function argument even when
2586 no method was removed.
2587 ** SHARED-INITIALIZE now initializes the values of the requested
2588 slots, including those with :ALLOCATION :CLASS.
2589 ** ALLOCATE-INSTANCE now works on structure classes defined via
2590 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2592 ** SLOT-EXISTS-P now works on conditions, as well as structures
2594 ** MAKE-LOAD-FORM now has the required methods on
2595 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2596 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2597 keyword, and now implements the SLOT-NAMES argument.
2598 ** methods with &OPTIONAL arguments no longer allow too many
2599 arguments to be passed in the call without error.
2600 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2601 option is consistent with the required arguments of the generic
2602 function lambda list.
2603 * bug fix: REQUIRE accepts a string designator. (Thanks to
2605 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2606 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2607 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2608 * bug fix: The compiler now checks for duplicated variables in macro
2610 * bug fix: SETQ on globals returns the correct value.
2611 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2612 not a COMPILER-ERROR (followed by some other strange error on
2613 choosing the CONTINUE restart).
2614 * bug fix: make.sh and friends are now more consistent in the way that
2615 they look for GNU "make".
2617 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2618 * experimental native threads support (on x86 Linux >=2.4 only).
2619 This is not compiled in by default: you need to add :SB-THREAD to
2620 the target features. See the "Beyond ANSI" chapter of the manual
2622 * fix for longstanding nonANSIism: The old distinction between
2623 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2624 The return value from CL:FIND-CLASS is now a CLOS class, and
2625 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2626 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2627 * An interface to the MetaObject Protocol, as described in Kiczales,
2628 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2629 MIT Press, 1991, is available from the SB-MOP package.
2630 * incompatible change: the SB-PCL package should now be considered
2631 a private implementation detail, and no longer a semi-private MOP
2633 * minor incompatible change: due to rearrangement for threads, the
2634 control stack and binding stack are now allocated at arbitrary
2635 addresses instead of being hardcoded per-port. Users affected by
2636 this probably have to be doing advanced things with shared
2637 libraries, and will know who they are.
2638 * minor incompatible change: Previously, all --eval forms used were
2639 processed with READ before any of them were processed with EVAL.
2640 Now each --eval form is processed with both READ and EVAL before
2641 the next --eval form is processed. (Thus package operations like
2642 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2643 work as the user might reasonably expect.)
2644 * minor incompatible change: *STANDARD-INPUT* is now only an
2645 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2647 * minor incompatible change: Y-OR-N-P is now character-oriented,
2648 not line oriented. Also, YES-OR-NO-P now works without errors.
2649 (thanks to Antonio Martinez)
2650 * sb-aclrepl module improvements: an integrated inspector, added
2651 repl features, and a bug fix to :trace command.
2652 * Known functions, which cannot be open coded by the backend, are
2653 considered to be able to check types of their arguments. (fixing
2654 a bug report by Nathan J. Froyd)
2655 * fixed a bug in computing method discriminating functions: It is
2656 now possible to define methods specialized on classes which have
2657 forward-referenced superclasses. (thanks to Gerd Moellmann)
2658 * fixed evaluation order in optional entries (reported by Gilbert
2660 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2661 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2662 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2663 required-by-AMOP NAME argument, as well as CLASS and
2664 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2665 * fixed bug 20: DEFMETHOD can define methods using names that are
2666 not the proper names of classes to designate class specializers.
2667 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2668 Unix file descriptors, instead of blowing up. (thanks to Antonio
2670 * Garbage collection refactoring: user-visible change is that a
2671 call to the GC function during WITHOUT-GCING will not do garbage
2672 collection until the end of the WITHOUT-GCING. If you were doing
2673 this you were probably losing anyway.
2674 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2675 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2676 TYPEP the latter but not the former.
2677 * The compiler issues a full WARNING for calls to undefined functions
2678 with names from the CL package.
2679 * MAP-INTO for a vector destination is open coded. (reported by
2680 Brian Downing on c.l.l)
2681 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2682 documentation string.
2683 * fixed some bugs revealed by Paul Dietz' test suite:
2684 ** COPY-ALIST now signals an error if its argument is a dotted
2686 ** Condition slots are now accessed more correctly in the presence
2687 of multiple initargs for a given slot.
2688 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2689 exclude restarts of the same name associated with a different
2691 ** DEFCLASS of forward-referenced classes with another
2692 forward-referenced class in the superclasses list no longer
2694 ** Condition slots are now initialized once each, not multiple
2695 times. (thanks to Gerd Moellmann)
2696 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2698 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2699 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2700 arguments contain duplicated elements.
2701 ** RESTART-CASE understands local macros.
2702 ** RESTART-CASE associates exactly its own restarts with a condition.
2703 ** ENDP in safe mode checks its argument to be of type LIST.
2704 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2705 in question is unbound.
2706 ** Optimized MAKE-INSTANCE functions no longer cause internal
2707 assertion failures in the presence of duplicate initargs.
2708 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2709 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2710 propagated even in the ctor optimized implementation of
2712 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2714 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2715 * a better implementation of SXHASH on (simple) bit vectors,
2716 measured both in execution speed and in distribution of results
2717 over the positive fixnums, has been installed. Likewise, a better
2718 implementation of EQUAL for simple bit vectors is now available.
2719 * fixed CEILING optimization for a divisor of form 2^k.
2720 * fixed bug 240 (emitting extra style warnings "using the lexical
2721 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2722 by Antonio Martinez)
2723 * fixed SXHASH, giving different results for NIL depending on type
2724 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2725 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2726 variable. (found by Rolf Wester)
2727 * SBCL does not ignore type declarations for special
2728 variables. (reported by rif on c.l.l 2003-03-05)
2729 * some bug fixes in contrib/sb-aclrepl/
2730 * fixed some bugs revealed by Paul Dietz' test suite:
2731 ** a bug in the CONS type specifier, whereby the CAR and CDR
2732 types got intertwined, has been fixed;
2733 ** the type system is now able to reason about the interaction
2734 between INTEGER and RATIO types more completely;
2735 ** APPEND, [N]REVERSE and NRECONC check that those their
2736 arguments, which must be proper lists, are really so;
2737 ** An array specialized to be unable to hold elements has been
2738 implemented, as required -- yes, really -- by ANSI;
2739 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2740 on malformed property lists;
2742 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2743 * incompatible packaging change: in line with Unix convention,
2744 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2745 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2746 * REQUIRE and PROVIDE are now optionally capable of doing something
2747 useful. See the documentation string for REQUIRE.
2748 * infrastructure for a managed SBCL contrib system: contributed
2749 modules in this release include:
2750 ** the ASDF system definition facility;
2751 ** an interface to the BSD Sockets API;
2752 ** an ACL-like convenience interface to the repl;
2753 (thanks to Kevin Rosenberg)
2754 ** an implementation of ROTATE-BYTE, with efficient implementation
2756 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2757 gives a closer approximation to the right answer than previously.
2758 (thanks to Raymond Toy)
2759 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2760 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2761 optional environment arguments, as required by ANSI.
2762 * fixed bugs in other functions taking environment objects, allowing
2763 calls with an explicit NIL environment argument to be compiled
2765 * fixed bug 228: primary return values from
2766 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2767 COMPILE or FUNCTION.
2768 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2769 :TYPE will no longer signal errors on innocuous objects.
2770 * fixed bug 231b: SETQ is better at respecting type declarations in
2771 the lexical environment.
2772 * fixed a bug in DEFCLASS: classes named by symbols with no or
2773 unprintable packages can now be defined.
2774 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2775 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2776 * fixed compiler failure related to checking types of functions.
2777 (reported by Robert E. Brown)
2778 * the compiler is now much more consistent in its error-checking
2779 treatment of bounding index arguments to sequence functions: in
2780 (SAFETY 3) code, errors will be signalled in almost all cases if
2781 invalid sequence bounding indices are passed to functions defined
2782 by ANSI to operate on sequences.
2783 * fixed a bug in the build procedure: documentation of SBCL-specific
2784 packages is now preserved and available in the final Lisp image.
2785 * lifted FDEFINITION lookup out of loops in the implementation of
2786 many list operations. (thanks to Robert E. Brown)
2787 * fixed a bug in the reader: the #n# reader macro now works for
2788 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2789 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2790 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2791 E. Brown and Tony Martinez respectively)
2792 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2793 SLOT-BOUNDP: in particular, it is now able to optimize them much
2794 better, and is now not vulnerable to having packages renamed.
2795 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2796 always called when a slot is not present in an instance. (thanks
2798 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2799 longer signal an error on unreading a character following EOF on
2800 the previous constituent stream. (thanks to Tony Martinez)
2801 * fixed some bugs revealed by Paul Dietz' test suite:
2802 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2803 not just nonnegative fixnums;
2804 ** the logical bit-array operators such as BIT-AND now accept an
2805 explicit NIL for their "opt-arg" argument (to indicate a
2806 freshly-consed result bit-array);
2807 ** ELT now signals an error on an invalid sequence index in safe
2809 ** the type system is now cleverer about negations of numeric
2810 types, and consequently understands the BIGNUM and RATIO types
2812 ** the type system is now cleverer about the interaction between
2813 INTEGER and RATIO types: while bugs still remain, many more
2814 cases are accurately computed;
2815 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2816 if it is in the last clause;
2817 ** CONSTANTLY now correctly returns a side-effect-free function in
2819 ** DECLARE is no longer treated as a special-operator; in
2820 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2821 * incremented fasl file version number due to the change in the
2822 DEFSTRUCT-SLOT-DESCRIPTION structure.
2824 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2825 * minor incompatible change: code processed by the "interpreter" or
2826 EVAL now has a compilation optimization policy of (DEBUG 2)
2827 (changed from (DEBUG 1)) to improve debuggability of interactive
2828 development, and to allow the use of the debug RETURN command in
2830 * an experimental implementation of the RETURN command for the
2831 debugger has been included. (thanks to Frederik Kuivinen)
2832 * fixed bug 62: constraints were not propagated into a loop.
2833 * fixed bug in embedded calls of SORT (reported and investigated by
2835 * fixed some bugs revealed by Paul Dietz' test suite:
2836 ** printing and reading of arrays with some dimensions having
2837 length 0 (thanks to Gerd Moellmann);
2838 ** BOA constructor with &AUX argument without a default value does
2839 not cause a type error;
2840 ** CONSTANTP now returns true for all self-evaluating objects.
2842 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2843 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2844 accessors that are related by inheritance, as specified in the
2845 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2846 Valtteri Vuorikoski)
2847 * The compiler is now able to inline functions that were defined in
2848 a complex lexical environment (e.g. inside a MACROLET).
2849 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2851 * Support for the upcoming FreeBSD-5.0 release has been included.
2852 (thanks to Dag-Erling Smorgrav)
2853 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2854 effect when it is not in a toplevel context.
2855 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2856 stand a better chance of being correct. (thanks to Gerd
2858 * fixed bug in COERCE, which now signals an error on coercing a
2859 rational to a bounded real type which excludes the expected
2861 * The compiler is now able to derive types more accurately from the
2862 COERCE and COMPILE functions.
2863 * fixed bug 223: functional binding is considered to be constant
2864 only for symbols in the CL package.
2865 * fixed bug 231: SETQ did not check the type of a variable being set
2866 (reported by Robert E. Brown)
2867 * A new optimization for MAKE-INSTANCE has been included, fixing
2868 various bugs (including relating to :ALLOCATION :CLASS slots and
2869 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2871 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2872 ** As required by ANSI, LOOP now disallows anonymous collection
2873 clauses such as COLLECT I in conjunction with aggregate boolean
2874 clauses such as THEREIS (= I 1);
2875 ** LOOP now signals an error when any variable is reused in the
2876 same loop (including the potentially useful construct analogous
2877 to WITH A = 1 WITH A = (1+ A);
2878 ** IT is only a special loop symbol within the first clause of a
2879 conditional loop clause;
2880 ** LOOP with a typed iteration variable over a hashtable now
2881 signals a type error iff it should.
2882 * fixed some other bugs revealed by Paul Dietz' test suite:
2883 ** FILE-STREAM now names the class previously known as FD-STREAM;
2884 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2885 argument) no longer signals an error;
2886 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2887 of the slot symbol, rather than using the current package
2888 ((:CONC-NAME "") continues to intern the slot's name in the
2890 * incremented fasl file version number, because of the incompatible
2891 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2892 of the new implementation of DEFINE-COMPILER-MACRO.
2894 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2895 * Support for building SBCL for MIPS platforms running in
2896 little-endian mode has now been checked in, and basic
2897 functionality on said platforms verified.
2898 * minor incompatible change: PCL now records the pathname of a file
2899 in which methods and the like are defined, rather than its
2901 * minor incompatible change: TRUENAME now considers the truename of
2902 a file naming a directory to be the pathname with :DIRECTORY
2903 component indicating that directory.
2904 * minor incompatible change: a NAMED clause in the extended form of
2905 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2906 reason for the previous behaviour is unclear.
2907 * more systematization and improvement of CLOS and MOP conformance
2908 in PCL (thanks to Gerd Moellman and Pierre Mai):
2909 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2911 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2912 primary methods with no specializers;
2913 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2915 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2916 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2917 FORWARD-REFERENCED-CLASSes; error reporting on
2918 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2920 ** SXHASH on CLOS instances now uses a slot internal to the
2921 instance to return different numbers on distinct instances,
2922 while preserving the same return value through invocations of
2924 ** DEFMETHOD signals errors when methods with longer incongruent
2925 lambda lists are added to generic functions;
2926 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2927 CLASS, as specified in AMOP;
2928 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2929 on the order returned by the primary method for classes of
2930 class STANDARD-CLASS;
2931 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2932 * fixed some bugs shown by Paul Dietz' test suite:
2933 ** DOLIST puts its body in TAGBODY;
2934 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2936 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2937 value producing form;
2938 ** if more variables are given to PROGV than values, extra
2939 variables are bound and made to have no value;
2940 ** NSUBSTITUTE on list arguments gets the right answer with
2942 ** ELT signals an error of type TYPE-ERROR when the index argument
2943 is not a valid sequence index;
2944 ** LOOP signals (at macroexpansion time) an error of type
2945 PROGRAM-ERROR when duplicate variable names are found;
2946 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2947 ** FUNCALL of special-operators now cause an error of type
2949 ** PSETQ now works as required in the presence of side-effecting
2950 symbol-macro places;
2951 ** NCONC accepts any object as its last argument;
2952 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2954 ** loop-for-as-package does not require a package to be explicitely
2956 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2958 * fixed bug 166: compiler preserves "there is a way to go"
2959 invariant when deleting code.
2960 * fixed bug 172: macro lambda lists with required arguments after
2961 &REST arguments now cause an error to be signalled. (thanks to
2963 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2964 bug. (thanks to Matthew Danish)
2965 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2967 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2968 arguments are lexically rebound. (thanks to Gerd Moellmann and
2970 * fixed bug 194: error messages are now more informative when there
2971 is no primary method applicable in a call to a generic
2972 function. (thanks to Gerd Moellmann)
2973 * fixed bug in command line argument checking (thanks to Julian
2975 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2976 (thanks to Matthew Danish)
2977 * incremented fasl file version number, because of the
2978 SXHASH-related changes in the layout of CLOS data structures
2980 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2981 * minor incompatible change: The runtime (the Unix executable named
2982 "sbcl") is now much pickier about the .core files it will load.
2983 Essentially it now requires .core files to descend from the same
2984 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2985 as the runtime does. (The intent is to prevent the crashes which
2986 can occur, and which can even be reported as mysterious failures,
2987 when people patch the sources or change the build parameters
2988 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2989 sbcl and .core files.)
2990 * fixed bug: VALUES-LIST is no longer optimized away.
2991 * fixed bug 142: The FFI conversion of C string values to Lisp
2992 string values no longer conses excessively. (thanks to Nathan
2993 Froyd porting Raymond Toy's fix to CMU CL)
2994 * began to systematize and improve MOP conformance in PCL (thanks to
2995 Nathan Froyd, Gerd Moellman and Pierre Mai):
2996 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2998 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2999 ** FINALIZE-INHERITANCE is now called on class finalization;
3000 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3001 argument precedence order.
3002 * fixed bug 202: The compiler no longer fails on functions whose
3003 derived types contradict their declared type.
3004 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3005 so it can be non-toplevel.
3006 * The fasl file version number has changed (because of the new
3007 implementation of DEFMACRO).
3008 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3009 safe code, that any length requirement by their type-specifier
3010 argument is valid. The exceptions to this are described in bug
3012 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3013 functions now check the types of their inputs as required by ANSI.
3014 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3015 introduced symbol is DECLAREd to be SPECIAL.
3016 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3017 * fixed bug: :COUNT argument to sequence functions may be negative
3018 * fixed bug: body of DO-SYMBOLS may contain declarations
3019 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3020 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3021 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3022 symbol macro only once
3023 * fixed printing of call frame when argument list is unavailable
3024 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3025 * compiler no longer signals WARNING on unknown keyword
3028 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3029 * A beta-quality port to the mips architecture running Linux,
3030 based on the old CMUCL backend, has been made. It has been tested
3031 on a big-endian kernel, and works sufficiently well to be able to
3032 rebuild itself; it has not been tested in little-endian mode.
3033 * fixed an inconsistency between gencgc.c and purify.c which made
3034 dumping/loading .core files unreliable
3035 * fixed bug 120a: The compiler now deals correctly with IFs where
3036 the consequent is the same as the alternative, instead of
3037 misderiving the return type. (thanks to Alexey Dejneka)
3038 * fixed bug 113: Logical pathnames are now dumpable (the logical
3039 host is resolved at load-time, throwing an error if it is not
3041 * fixed bug 174: FORMAT's error message is slightly clearer when a
3042 non-printing character is used in a format directive.
3043 * fixed several bugs in compiler checking of type declarations, i.e.
3044 violations of the Python "declarations are assertions" principle
3045 (thanks to Alexey Dejneka)
3046 * fixed several bugs in PCL's error checking (thanks to Gerd
3048 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3050 * fixed bug in compilation of functions as first class values
3051 (thanks to Antonio Martinez-Shotton)
3052 * The compiler's handling TYPE-ERRORs which it can prove will
3053 inevitably happen at runtime has been cleaned up and corrected
3054 in several ways. (thanks to Alexey Dejneka)
3055 * improved argument type checking for various basic arithmetic
3056 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3057 lavished on them in the past that they can be compiled in many
3058 ways in different special cases
3059 * fixed bug 181: compiler checks validity of user supplied type
3061 * cleaned up code flushing in optimization: Function calls which
3062 should signal errors for safety purposes (e.g. which ANSI says
3063 should signal errors when their arguments are of incorrect type)
3064 are no longer optimized away.
3065 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3066 * incremented fasl file version number, because changes in the
3067 implementation of sequence functions like COERCE caused
3068 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3069 in old inline expansions) to become undefined. (Actually these
3070 changes were later undone, so we might very well be binary
3071 compatible with 0.7.7 after all, but leaving the version number
3072 incremented seemed like the simplest and most conservative
3075 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3076 * An alpha-quality port to the parisc architecture running Linux,
3077 based on the old CMUCL backend, has been made. This, even more so
3078 than the other backends, should be considered still a work in
3079 progress; known problems include that the Linux kernel in 64-bit
3080 mode does not propagate the correct sigcontext structure to
3081 userspace, and consequently SBCL on a parisc64 kernel will not
3083 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3084 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3085 suppressed.) Also now that the compiler is looking at declarations
3086 in the environment, it checks optimization declarations as well,
3087 and suppresses inlining when (> DEBUG SPEED).
3088 * More fixes have been made to treatment of floating point exception
3089 treatment and other Unix signals. In particular, floating point
3090 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3091 * The detection and handling of control stack exhaustion (infinite
3092 or very deeply nested recursion) has changed. Stack exhaustion
3093 detection is now done by write-protecting pages at the OS level
3094 and applies at all optimization settings; when found, a
3095 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3096 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3097 be caught using IGNORE-ERRORS.
3098 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3099 correctly when transforming recursive LABELS functions to LETs.
3100 (thanks to Alexey Dejneka)
3101 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3102 that are names of constants or global variables.
3103 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3104 alien routines with docstrings.
3105 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3106 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3108 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3109 (thanks to Eric Marsden)
3110 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3111 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3112 * Bug fix: The SPARC backend can now compile functions involving
3113 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3114 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3115 to a functions expecting a PATHNAME-DESIGNATOR.
3116 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3117 lambda lists. (thanks to Alexey Dejneka)
3118 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3119 functions so that EXPORTing the name of the class doesn't cause
3120 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3121 to Antonio Martinez for reporting this)
3122 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3123 constant-folded without failing an assertion. (thanks to Einar
3124 Floystad Dorum for reporting this)
3125 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3126 types are now dealt with more correctly by the compiler.
3127 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3128 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3129 behaviour analogous to RENAME-FILE. This puts its behaviour more
3130 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3132 * The fasl file version number has changed again. (because of the
3133 bug fix involving the names of PCL MAKE-INSTANCE functions)
3135 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3136 * bug fix: Floating point exceptions are treated much more
3137 consistently on the x86/Linux and PPC/Linux platforms.
3138 * Array initialization with :INITIAL-ELEMENT is now much faster for
3139 cases when the compiler cannot open code the array creation, but
3140 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3141 array accesses have also seen a speed increase.
3142 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3143 specified. (This was at the root of some bad interactions between
3144 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3146 * bug fix: Internal error arguments for undefined functions are now
3147 computed correctly on the PPC/Linux platform.
3148 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3149 Raymond Toy's patch for CMU CL)
3150 * Support for the Solaris 9 operating environment has been included
3151 (thanks to Daniel Merritt)
3152 * A very ugly but hopefully complete draft of the missing FFI chapter
3153 of the manual has been created by reformatting the corresponding
3154 CMU CL manual chapter into (currently very ugly and incoherent)
3155 DocBook and bringing it up to date for SBCL behavior. Thus, the
3156 manual is now essentially complete, at least by my extreme
3157 once-and-only-once standards, whereby it's acceptable to refer to
3158 the doc strings of SB-EXT functions as the primary documentation.
3159 * The fasl file version number has changed again, due to cleanup of
3160 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3161 is no longer a static symbol.)
3163 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3164 * SBCL now builds with OpenMCL (version 0.12) as the
3165 cross-compilation host; also, more progress has been made toward
3166 bootstrapping under CLISP.
3167 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3169 * bug 158 fixed: The compiler can now deal with integer loop
3170 increments different from 1; fixing this turned out also to fix
3172 * bug 169 fixed: no more bogus warnings about using lexical bindings
3173 despite the presence of perfectly good SPECIAL declarations (thanks
3174 to David Lichteblau)
3175 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3176 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3177 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3179 * bug 180 fixed: Method combination specifications no longer ignore
3180 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3181 * bug fix: Structure type predicate functions now check their argument
3182 count as they should.
3183 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3184 correctly. (thanks to Pierre Mai)
3185 * minor incompatible change: The --noprogrammer option is deprecated
3186 in favor of the new --disable-debugger option, which is very similar.
3187 (The major difference is that it takes effect at a slightly different
3188 time at startup, causing handling of errors in --sysinit and
3189 --userinit files will be affected differently.) The
3190 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3191 been added to allow this functionality to be controlled from ordinary
3192 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3193 maintainers, who might want to run non-interactive scripts to
3194 build SBCL cores which will later be used interactively.)
3195 * minor incompatible change: The LOAD function no longer, when given
3196 a wild pathname to load, loads all files matching that pathname.
3197 Instead, an error of type FILE-ERROR is signalled.
3199 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3200 * bug 147 fixed: The compiler preserves its block link/count
3201 invariants more correctly now so that it doesn't crash. (thanks
3203 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3205 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3206 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3207 no-value-for-this-character as NIL (as specified by ANSI).
3208 * HOST-NAMESTRING on physical pathnames now returns a string that is
3209 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3210 (thanks to Christophe Rhodes)
3211 * The Alpha port handles icache flushing more correctly. (thanks to
3213 * More progress has been made toward bootstrapping under CLISP. (thanks
3214 to Christophe Rhodes)
3215 * The fasl file format has changed again, because dynamic loading
3216 on OpenBSD (which has non-ELF object files) motivated some cleanups
3217 in the way that foreign symbols are transformed and passed around.
3218 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3219 is no longer treated as whitespace by the reader, but instead as
3220 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3221 |A
\7fB|, instead of A as it used to.
3223 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3224 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3225 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3226 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3227 this as of 0.7.1.45, but was left out of the previous news section
3228 (thanks to Dan Barlow)
3229 * SBCL now runs on the Solaris operating system on SPARC architectures
3230 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3231 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3232 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3233 * SPARC backend cleanups, allowing builds of cores optimized for V8
3234 and V9 SPARCS, and also emission of code targeted to a particular
3235 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3237 * SBCL is closer to bootstrapping under CLISP, thanks to various
3238 fixes by Christophe Rhodes.
3239 * The fasl file format has changed again, to allow the compiler's
3240 INFO database to support symbol macros.
3241 * The user manual (in doc/) is formatted into HTML more nicely.
3242 (thanks to coreythomas)
3243 * The system is smarter about SUBTYPEP relationships, especially
3244 those involving NOT types (including types such as ATOM which are
3245 represented internally using NOT types). Thus SUBTYPEP is less
3246 likely to return (VALUES NIL NIL) in general, and in particular
3247 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3249 * The fasl file format has changed again, because the internal
3250 representation of types now includes a new slot to support the new
3251 SUBTYPEP-of-NOT-types logic.
3252 * (not a change in the main branch of SBCL, but a related prototype
3253 which can hopefully be merged into the main branch of SBCL in the
3254 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3255 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3256 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3257 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3258 dispatch macro character combinations. (thanks to Alexey Dejneka)
3259 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3260 that superficially look like logical namestrings correctly.
3261 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3263 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3264 default supplied-p) for &optional and &key arguments. (thanks to
3267 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3268 * incompatible change: The compiler is now less aggressive about
3269 tail call optimization, doing it only when (> SPACE DEBUG) or
3270 (> SPEED DEBUG). (This is an incompatible change because there are
3271 programs which relied on the old CMU-CL-style behavior to optimize
3272 away their unbounded recursion which will now die of stack overflow.)
3273 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3274 for non-GENCGC systems has been increased to 20M (since that
3275 seems much closer to the likely performance optimum for modern
3276 systems than the old 4M value was)
3277 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3278 and *DEBUG-PRINT-LEVEL*
3279 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3280 of CMU CL's support for SPARC, and various endianness and other
3281 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3282 * new syntactic sugar for the Unix command line: --load foo.bar is now
3283 an alternate notation for --eval '(load "foo.bar")'.
3285 ** The system now detects stack overflow and handles it gracefully,
3286 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3287 optimization settings. (This is a good thing in general, and
3288 its introduction in this version should be particularly timely
3289 for anyone whose code fails because of suppression of tail
3291 ** The system now hunts for the C variable "environ" in a more
3292 devious way, to avoid segfaults when the C library version differs
3293 between compile time and run time. (thanks to Christophe Rhodes)
3294 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3295 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3296 ** The compiler no longer issues bogus style warnings for undefined
3297 classes in the same source file as the DEFCLASSes which defined
3298 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3300 ** fixes in CONDITION class precedence list for undefined function
3301 errors (thanks to Alexei Dejneka)
3302 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3303 correctly. (thanks to Dan Barlow)
3304 ** portability fixes aiming at bootstrapping under CLISP (thanks
3305 to Dave McDonald and Christophe Rhodes)
3306 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3307 ** fixes in type translation and and type inference (thanks to
3309 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3310 ** various fixes in the new ports (thanks to Dan Barlow)
3311 * several changes related to debugging:
3312 ** suppression of tail recursion, as noted above
3313 ** stack overflow detection, as noted above
3314 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3315 is now the default. (For some time encapsulation has been more
3316 reliable than the breakpoint-based :ENCAPSULATE NIL
3317 implementation, at least on X86 systems; and I just noticed that
3318 encapsulation also seems closer to the spirit of the ANSI
3321 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3323 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3324 up properly again. (There was a packaging bug in 0.7.0 which
3325 left their definitions in SB-SYS::LOAD-FOREIGN and
3326 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3327 vital for most things which interface to C-level interfaces,
3328 like extensions working with sockets or databases or
3329 Perl-compatible regexes or whatever, and the need to fix
3330 this bug is the main reason that 0.7.1 was released so
3332 ** DEFGENERIC is now choosier about the methods it redefines, so that
3333 reLOADing a previously-LOADed file containing DEFGENERICs does
3334 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3335 cycle now works as it should. (thanks to Alexey Dejneka)
3336 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3337 fixed by Christophe Rhodes. (He actually submitted this patch
3338 months ago, and I delayed until after 0.7.0.)
3339 ** Bug 111 (internal compiler confusion about runtime checks on
3340 FUNCTION types) was fixed by Alexey Dejneka.
3341 * Some internal cleanups (getting rid of variables which aren't
3342 needed now that the byte interpreter is gone) caused the fasl
3343 file format number to change again.
3345 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3346 * major incompatible change: The default fasl file extension, i.e. the
3347 default extension for files produced by COMPILE-FILE, has changed
3348 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3350 ** There are many changes in the implementation of the compiler.
3351 SBCL is now essentially a compiler-only implementation of ANSI
3352 Common Lisp. EVAL still "interprets" a few special cases, but
3353 almost all the interesting cases are handled by creating
3354 a LAMBDA expression, calling COMPILE on it, then calling
3355 FUNCALL on the result.
3356 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3357 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3358 Since the code is newer, there might still be some new bugs
3359 (though not as many as before Martin Atzmueller's fixes:-). But
3360 the new code is substantially simpler and clearer, and hopefully
3361 any remaining bugs will be simpler, less fundamental, and more
3362 fixable then the bugs in the old code.
3363 ** The revised compiler is still a little unsteady on its feet.
3365 *** The debugging information it produces (particularly the names
3366 of FUNCTION objects) is sometimes much less useful than what
3367 the old compiler produced.
3368 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3369 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3370 within a MACROLET) has been temporarily weakened.
3371 ** There are new compiler optimizations for various functions:
3372 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3373 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3374 *** the math functions TRUNCATE, FLOOR, and CEILING
3375 *** the function-of-all-trades COERCE
3376 Mostly these should be transparent, but there's one
3377 potentially-annoying problem (bug 117): when the compiler
3378 inline-expands a function and does type analysis on the result,
3379 it can create control paths which have type mismatches, and
3380 when it can't prove that those control paths aren't taken,
3381 it will issue WARNINGs about the type mismatches. This is
3382 a particular problem in practice for the new sequence functions.
3383 It's not clear how this should be fixed, and for now, a
3384 workaround is given in the entry for 117 in the BUGS file.
3385 ** (Because of the interaction between the two previous items --
3386 occasional inlining problems and new inline expansions -- some
3387 of the new sequence function optimizations won't really kick in
3388 completely until debugging information, and then inlining, are
3389 straightened out in some future version.)
3390 * minor incompatible changes:
3391 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3392 DIRECTORY no longer implicitly promotes NIL slots of its
3393 pathname argument to :WILD. In particular, when you ask for the
3394 contents of a directory (which you used to be able to do without
3395 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3396 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3397 ** changes in behavior that ANSI explicitly defines to be
3398 implementation dependent:
3399 *** The new compiler-only implementation still conforms with ANSI,
3400 but acts a little different than before. Besides the obvious
3401 changes in performance tradeoffs (that the cost per form passed
3402 to EVAL has gone up, and the cost per form executed by EVAL
3403 has gone down), the behavior of the system changes a little
3404 because there are no longer any interpreted function objects.
3405 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3406 e.g. doing COMPILE on the output of interactive DEFUN is
3408 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3410 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3411 has changed from (:CL) to NIL.
3412 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3413 e.g. #\Newline and #\Tab, don't have names specified in the
3414 ANSI Common Lisp standard, is now based on their ASCII symbolic
3415 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3416 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3417 are no longer used for output.
3418 ** changes in internal implementation constants:
3419 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3420 4 million. (If your application spends a lot of time GCing and
3421 you have a lot of RAM, you might want to experiment with
3422 increasing it even more.)
3423 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3424 However, almost all old code should still continue to work without
3425 immediate update, as SB-C-CALL is now a (deprecated) nickname
3427 ** Old operator names in the style DEF-FOO are now deprecated in
3428 favor of new corresponding names DEFINE-FOO, for consistency with
3429 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3430 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3431 internal symbols, but a few supported extensions like
3432 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3433 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3434 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3435 etc. as you get deeper into recursive calls to the debugger
3436 command loop, instead of the old "5]", "5]]", "5]]]"
3437 sequence. (I was motivated to do this when squabbles between
3438 ILISP and SBCL left me very deeply nested in the debugger. In the
3439 short term, this change will probably provoke more ILISP/SBCL
3440 squabbles, but hopefully it will be an improvement in the long run.)
3441 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3442 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3443 function, since declaiming return types involving aliens is
3444 (1) annoyingly messy to do by hand and (2) vital to efficient
3445 compilation of code which calls such functions.
3446 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3447 longer reexported by the SB-EXT package. They're solely useful
3448 for alien code, so it seems more logical that you should get
3449 them from the SB-ALIEN package, not in SB-EXT.
3450 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3451 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3452 features. Instead, the code that they used to control is always
3453 built into the system.
3454 * many other bug fixes
3455 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3456 advantage of the new EVAL-WHEN stuff and to clean them up in
3457 general, and they are now more ANSI-compliant in a number of
3458 ways. Martin Atzmueller is responsible for a lot of this.
3459 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3461 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3462 *** correct ERROR type for various file operations
3463 *** some fixes for Lisp streams
3464 *** DEFMETHOD syntax checking
3465 *** changing old weird representation of debug information as
3466 strings (which, among their other deficiencies, don't transform
3467 correctly when you rename packages, and don't change their
3468 print representation when you change things like *PACKAGE*
3469 and *PRINT-LENGTH*) to symbols and lists of symbols
3470 He also made several improvements and fixed several bugs in DESCRIBE.
3471 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3473 *** misbehavior of WRITE-STRING/WRITE-LINE
3474 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3475 and several other LOOP problems as well
3476 *** DIRECTORY when similar filenames are present
3477 *** DEFGENERIC with :METHOD options
3478 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3479 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3480 *** argument ordering in FIND with :TEST option
3481 *** mishandled package designator argument in APROPOS-LIST
3482 *** various problems in the backquote readmacro
3483 *** a bug in APROPOS
3484 *** probably some others that I'm not describing very well here,
3485 since the CVS log documents them by reference to sbcl-devel
3486 messages, and the SourceForge archives aren't working well.:-(
3487 ** Dan Barlow improved the Alpha port (and is making progress on the
3488 PPC port, for those of you who think different).
3489 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3490 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3491 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3492 continues to work on the SPARC port (for those of you in a position
3493 to look down upon our little PC-compatible boxes from a great height).
3494 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3495 and uses that copy, rather than the current dynamic value, when
3496 it's trying to decide whether to truncate output. Thus e.g.
3497 (let ((*print-lines* 50))
3498 (pprint-logical-block (stream nil)
3500 (let ((*print-lines* 8))
3501 (print (aref possiblybigthings i) stream)))))
3502 should now truncate the logical block only at 50 lines, instead of
3503 often truncating it at 8 lines, as it did before.
3504 * The doc/cmucl/ directory, containing old CMU CL documentation
3505 from the time of the fork, is no longer part of the base system.
3506 SourceForge has shut down its anonymous FTP service, and with it
3507 my original plan for distributing the old CMU CL documentation
3508 there. For now, if you need these files you can download an old
3509 SBCL source release and extract them from it.
3510 * The fasl file version number changed again, for dozens of reasons,
3511 some of which are apparent above.
3513 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3514 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3515 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3516 patch, so that DISASSEMBLE output is much nicer.
3517 * The code in the SB-PROFILE package now seems reasonably stable.
3518 I still haven't decided what the final interface should look like
3519 (I'd like PROFILE to interact cleanly with TRACE, since both
3520 facilities use function encapsulation) but if you have a need
3521 for profiling now, you can probably use it successfully with
3522 the current CMU-CL-style interface.
3523 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3524 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3525 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3526 link now returns the dangling link itself, and for similar
3527 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3528 link itself. (In these cases the old code signalled an error and
3529 looped endlessly, respectively.) Thus, DIRECTORY now works even
3530 in the presence of dangling and cyclic symbolic links.
3531 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3532 is now a supported extension again, since the consensus on
3533 sbcl-devel was that it can be useful for ordinary development
3534 work, not just for debugging SBCL itself.
3535 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3536 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3537 that currently-defined functions might be redefined later with
3538 different return types.
3539 * Hash tables can be printed readably, as inspired by CMU CL code
3540 of Eric Marsden and SBCL code of Martin Atzmueller.
3541 * better error handling in CLOS method combination, thanks to
3542 Martin Atzmueller porting Pierre Mai's CMU CL patches
3543 * more overflow fixes for >16Mbyte I/O buffers
3544 * A bug in READ has been fixed, so that now a single Ctrl-D
3545 character suffices to cause end-of-file on character streams.
3546 In particular, now you only need one Ctrl-D at the command
3547 line (not two) to exit SBCL.
3548 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3550 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3551 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3552 does the right thing.
3553 * The implementation of some type tests, especially for CONDITION
3554 types, is now tidier and maybe faster, due to CMU CL code
3555 originally by Douglas Crosher, ported by Martin Atzmueller.
3556 * Some math functions have been fixed, and there are new
3557 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3558 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3559 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3560 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3561 likely to be built into sbcl-0.7.0.)
3562 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3563 CL, ported to SBCL by Martin Atzmueller)
3564 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3565 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3566 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3567 * There's a new slam.sh hack to shorten the edit/compile/debug
3568 cycle for low-level changes to SBCL itself, and a new
3569 :SB-AFTER-XC-CORE target feature to control the generation of
3570 the after-xc.core file needed by slam.sh.
3571 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3572 COMPILE-FILE is no longer supported, so that now every function
3573 gets an entry point, so that block compilation looks a little
3574 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3575 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3576 returns the number of bytes consed since the system started,
3577 rather than the number consed since the first time the function
3578 was called. (The new definition parallels ANSI functions like
3579 CL:GET-INTERNAL-RUN-TIME.)
3580 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3581 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3582 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3583 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3584 * Fasl file version numbers are now independent of the target CPU,
3585 since historically most system changes which required version
3586 number changes have affected all CPUs equally. Similarly,
3587 the byte fasl file version is now equal to the ordinary
3590 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3591 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3592 is no longer recognized. I apologize for this, because it was
3593 listed in SB-EXT as a supported extension, but I found that
3594 its existing behavior was poorly specified, as well as incorrectly
3595 specified, and it looked like too much of a mess to straighten it
3596 out. I have enough on my hands trying to get ANSI stuff to work..
3597 * many patches ported from CMU CL by Martin Atzmueller, with
3598 half a dozen bug fixes in pretty-printing and the debugger, and
3599 half a dozen others elsewhere
3600 * fixed bug 13: Floating point infinities are now supported again.
3601 They might still be a little bit flaky, but thanks to bug reports
3602 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3603 as flaky as they were.
3604 * The --noprogrammer command line option is now supported. (Its
3605 behavior is slightly different in detail from what the old man
3606 page claimed it would do, but it's still appropriate under the
3607 same circumstances that the man page talks about.)
3608 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3609 are now supported, and enabled by default. Thus, the compiler can
3610 handle many floating point and complex operations much less
3611 inefficiently. (Thus e.g. you can implement a complex FFT
3613 * The compiler now detects type mismatches between DECLAIM FTYPE
3614 and DEFUN better, and implements CHECK-TYPE more correctly, and
3615 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3616 thanks to patches from Martin Atzmueller.
3617 * various fixes to make the cross-compiler more portable to
3618 ANSI-conforming-but-different cross-compilation hosts (notably
3619 Lispworks for Windows, following bug reports from Arthur Lemmens)
3620 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3621 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3622 fixed thanks to Pierre Mai's CMU CL patches.
3623 * fixes to keep the system from overflowing internal counters when
3624 it tries to use i/o buffers larger than 16M bytes
3625 * fixed bug 45a: Various internal functions required to support
3626 complex special functions have been merged from CMU CL sources.
3627 (When I was first setting up SBCL, I misunderstood a compile-time
3628 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3629 * improved support for type intersection and union, fixing bug 12
3630 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3631 more obscure bugs as well
3632 * some steps toward byte-compiling non-performance-critical
3633 parts of the system, courtesy of patches from Martin Atzmueller
3634 * Christophe Rhodes has made some debian packages of sbcl at
3635 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3636 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3637 stable, but are nonetheless usable. When he's ready, I'd be happy
3638 to add them to the SourceForge "File Releases" section. (And if
3639 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3640 * new fasl file format version number (because of changes in
3641 internal representation of (OR ..) types to accommodate the new
3642 support for (AND ..) types, among other things)
3644 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3645 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3647 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3649 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3650 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3651 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3652 * DESCRIBE now works on structure objects again.
3653 * Most function call argument type mismatches are now handled as
3654 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3655 know whether the function will be redefined before the call is
3656 executed. (The compiler could flag local calls with full WARNINGs,
3657 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3658 it doesn't keep track of enough information to know whether calls
3659 are local in this sense.)
3660 * Compiler output is now more verbose, with messages truncated
3661 later than before. (There should be some supported way for users
3662 to override the default verbosity, but I haven't decided how to
3663 provide it yet, so this behavior is still controlled by the internal
3664 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3665 src/compiler/ir1util.lisp.)
3666 * Fasl file format version numbers have increased again, because
3667 support for the Gray streams extension changes the layout of the
3668 system's STREAM objects.
3669 * The Gray subclassable streams extension now works, thanks to a
3670 patch from Martin Atzmueller.
3671 * The full LOAD-FOREIGN extension (not just the primitive
3672 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3673 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3674 but like most other programs, it defaults to copying the Unix
3675 environment from the original process instead of starting the
3676 new process in an empty environment.
3677 * Extensions which manipulate the Unix environment now support
3678 an :ENVIRONMENT keyword option which doesn't smash case or
3679 do other bad things. The CMU-CL-style :ENV option is retained
3680 for porting convenience.
3681 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3682 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3684 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3686 * A patch from Martin Atzmueller seems to have solved the SIGINT
3687 problem, and as far as we know, signal-handling now works cleanly.
3688 (If you find any new bugs, please report them!)
3689 * The system no longer defaults Lisp source file names to types
3690 ".l", ".cl", or ".lsp", but only to ".lisp".
3691 * The compiler no longer uses special default file extensions for
3692 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3693 seems to expect a single default extension for all compiled code,
3694 and there's no compelling reason to try to stretch the standard
3695 to allow two different extensions.) Instead, byte-compiled files
3696 default to the same extension as native-compiled files.
3697 * Fasl file format version numbers have increased again, because
3698 a rearrangement of internal implementation packages made some
3699 dumped symbols in old fasl files unreadable in new cores.
3700 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3701 many fewer weird special cases.
3702 * Bug #17 (differing COMPILE-FILE behavior between logical and
3703 physical pathnames) has been fixed, and some related misbehavior too,
3704 thanks to a patch from Martin Atzmueller.
3705 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3706 by Tim Moore, ported to SBCL by Martin Atzmueller.
3707 * Martin Atzmueller fixed several filesystem-related problems,
3708 including bug #36, in part by porting CMU CL patches, which were
3709 written in part by Paul Werkowski.
3710 * More compiler warnings in src/runtime/ are gone, thanks to
3711 more patches from Martin Atzmueller.
3712 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3715 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3717 * DESCRIBE now works on CONDITION objects.
3718 * The debugger now handles errors which arise when trying to print
3719 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3721 * The build system now uses an additional file, customize-target-features.lisp,
3722 to allow local modifications to the target *FEATURES* list. (The point of
3723 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3724 debugging features enabled, without having to worry about propagating it
3725 into everyone's system when I do a "cvs update".) When no
3726 customize-target-features.lisp file exists, the target *FEATURES* list
3727 should be constructed the same way as before.
3728 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3729 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3730 warn about behavior which is undefined under the ANSI spec. Note: This
3731 is specified by ANSI, but it's not very popular with programmers.
3732 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3733 macro in the SBCL sources for an example of a workaround which you
3734 might use to make portable ANSI-standard code which does what you want.
3735 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3736 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3737 when it's used as a non-toplevel form. (This is required in order
3738 to implement the DEFCONSTANT-EQX macro.)
3739 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3740 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3741 funny things, due to bugs in the implementation of EVAL-WHEN
3742 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3743 non-toplevel DEFCONSTANTs are uncommon.)
3744 * The core file version number and fasl file version number have been
3745 incremented, because the old noncompliant DEFCONSTANT behavior involved
3746 calling functions which no longer exist, and because I also took the
3747 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3748 * fixed bug 1 (error handling before read-eval-print loop starts), and
3749 redid debugger restarts and related debugger commands somewhat while
3751 ** The QUIT debugger command is gone, since it did something
3752 rather different than the SB-EXT:QUIT command, and since it never
3753 worked properly outside the main toplevel read/eval/print loop.
3754 Invoking the new TOPLEVEL restart provides the same functionality.
3755 ** The GO debugger command is also gone, since you can just invoke
3756 the CONTINUE restart directly instead.
3757 ** The TOP debugger command is also gone, since it's redundant with the
3758 FRAME 0 command, and since it interfered with abbreviations for the
3760 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3762 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3763 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3764 some porting work by Martin Atzmueller)
3765 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3766 with a boatload of patches which he ported to SBCL. Now that those
3768 ** The system tries to make sure that its low-priority messages
3769 are prefixed by semicolons, to help people who like to use
3770 syntax highlighting in their ILISP buffer. (This patch
3771 was originally due to Raymond Toy.)
3772 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3773 patches originally written by Raymond Toy.
3774 ** The compiler understands coercion between single-value and
3775 multiple-VALUES type expressions better, getting rid of some very
3776 weird behavior, thanks to patches originally by Robert MacLachlan
3777 and Douglas Crosher.
3778 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3779 lambda lists, thanks to a patch originally by Pierre Mai.
3780 ** The system no longer bogusly warns about "abbreviated type
3782 ** The compiler gets less confused by inlining and RETURN-FROM,
3783 thanks to some patches originally by Tim Moore.
3784 ** The system no longer hangs when dumping circular lists to fasl
3785 files, thanks to a patch originally from Douglas Crosher.
3786 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3787 undefined function error.
3788 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3789 in structure slot initforms) for now, documented workaround instead:-|
3790 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3791 * fixed bug 5: added stubs for various Gray stream functions called
3792 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3793 installed, at least appropriate type errors are generated
3794 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3795 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3796 consistently in DEFMETHOD forms.
3797 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3798 it doesn't seem to affect SBCL after all
3799 * The C runtime system now builds with better optimization and many
3800 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3802 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3804 * The system is now under CVS at SourceForge (instead of the
3805 CVS repository on my home machine).
3806 * The new signal handling code has been tweaked to treat register
3807 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3808 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3809 downstream, e.g. in debug-int.lisp, has implicit dependencies
3810 on the unsignedness of integer representation of machine words,
3811 and that caused the system to bomb out with infinite regress
3812 when trying to recover from type errors involving signed values,
3813 e.g. (BUTLAST '(1 2 3) -1).)
3814 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3815 (This was one of the bugs Peter Van Eynde reported back in July.)
3816 * The system now uses code inspired by Colin Walters' O(N)
3817 implementation of MAP (from the cmucl-imp@cons.org mailing
3818 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3819 inline the MAP operation, and there is more than one
3820 sequence argument to the MAP call (so that it can't just
3821 do ETYPECASE once and for all based on the type of the
3822 single sequence argument). (The old non-inline implementation
3823 of the general M-argument sequence-of-length-N case required
3824 O(M*N*N) time when any of the sequence arguments were LISTs.)
3825 * The QUIT :UNIX-CODE keyword argument has been renamed to
3826 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3827 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3829 * Raymond Wiker's patches to port dynamic loading from Linux to
3830 FreeBSD have been added.
3831 * The BUGS file is now more nearly up to date, thanks in large part
3832 to Martin Atzmueller's review of it.
3833 * The debugger now flushes standard output streams before it begins
3834 its output ("debugger invoked" and so forth).
3835 * The core version number and fasl file version number have both
3836 been incremented, because of incompatible changes in the layout
3838 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3839 used to not be called for a saved Lisp image.)
3840 * Martin Atzmueller's version of a patch to fix a compiler crash,
3841 as posted on sbcl-devel 13 September 2000, has been installed.
3842 * Instead of installing Martin Atzmueller's patch for the
3843 compiler transform for SUBSEQ, I deleted the compiler transform,
3844 and transforms for some similar consing operations.
3845 * A bug in signal handling which kept TRACE from working on OpenBSD
3847 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3848 away by constant folding
3849 * The system now defines its address space constants in one place
3850 (in the Lisp sources), and propagates them automatically elsewhere
3851 (through GENESIS and the sbcl.h file). Therefore, patching the
3852 address map is less unnecessarily tedious and error-prone. The
3853 Lisp names of address space constants have also been systematized.
3854 * CVS tags like dollar-Header-dollar have been removed from
3855 the sources, because they have never saved me trouble and
3856 they've been source of trouble working with patches and other
3857 diff-related operations.
3858 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3859 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3861 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3863 * The system has been ported to OpenBSD.
3864 * The system now compiles with a simple "sh make.sh" on the systems
3865 that it's supported on. I.e., now you no longer need to tweak
3866 text in the target-features.lisp-expr and symlinks in src/runtime/
3867 by hand, the make.sh takes care of it for you.
3868 * The system is no longer so grossly inefficient when compiling code
3869 involving vectors implemented as general (not simple) vectors (VECTOR T),
3870 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3871 which dares to use the various sequence functions on non-simple
3872 vectors, takes less of a performance hit.
3873 * There is now a primitive type predicate VECTOR-T-P
3874 to test for the (VECTOR T) type, so that e.g.
3875 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3876 can now be compiled with some semblance of efficiency. (The old code
3877 turned the type declaration into a full call to %TYPEP at runtime!)
3878 * AREF on (VECTOR T) is still not fast, since it's still compiled
3879 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3880 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3881 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3882 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3883 without making full calls to SUBTYPEP at runtime.
3884 (Some analogous efficiency issues for non-simple vectors specialized to
3885 element types other than T, or for non-simple multidimensional arrays,
3886 have not been addressed. They could almost certainly be handled the
3887 same way if anyone is motivated to do so.)
3888 * The changes in array handling break binary compatibility, so
3889 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3890 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3891 instead of (VALUES T T).
3892 * By following the instructions that Dan Barlow posted to sbcl-devel
3893 on 2 July 2000, I was able to enable primitive dynamic object
3894 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3895 functionality is not implemented (since it calls ld to resolve
3896 library references automatically, requiring RUN-PROGRAM for its
3897 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3898 try to resolve library references) is now supported.
3899 * The system now flushes the standard output streams when it terminates,
3900 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3901 them at several other probably-convenient times, e.g. in each pass of
3902 the toplevel read-eval-print loop, and after evaluating a form given
3903 as an "--eval" command-line option. (These changes were motivated by a
3904 discussion of stream flushing issues on cmucl-imp in August 2000.)
3905 * The source transform for TYPEP of array types no longer assumes
3906 that an array whose element type is a not-yet-defined type
3907 is implemented as an array of T, but instead punts, so that the
3908 type will be interpreted at runtime.
3909 * There is now some support for cross-compiling in make.sh: each of
3910 the phases of make.sh has its own script. (This should be transparent
3911 to people doing ordinary, non-cross-compile builds.)
3912 * Since my laptop doesn't have hundreds of megabytes of memory like
3913 my desktop machine, I became more motivated to do some items on
3914 my to-do list in order to reduce the size of the system a little:
3915 ** Arrange for various needed-only-at-cold-init things to be
3916 uninterned after cold init. To support this, those things have
3917 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3918 symbols with such names are now uninterned after cold init).
3919 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3920 fasl files for cold load.
3921 ** Remove the old compiler structure pooling code (which used to
3922 be conditional on the target feature :SB-ALLOC) completely.
3923 ** Redo the representation of some data in cold init to be more compact.
3924 (I also looked into supporting byte compiled code at bootstrap time,
3925 which would probably reduce the size of the system a lot, but that
3926 looked too complicated, so I punted for now.)
3927 * The maximum signal nesting depth in the src/runtime/ support code has
3928 been reduced from 4096 to 256. (I don't know any reason for the very
3929 large old value. If the new smaller value turns out to break something,
3930 I'll probably just bump it back up.)
3931 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3933 * Many, many bugs reported by Peter Van Eynde have been added to
3934 the BUGS list; some have even been fixed.
3935 * While enabling dynamic object file loading, I tried to make the
3936 code easier to understand, renaming various functions and variables
3937 with less ambiguous names, and changing some function calling
3938 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3939 * While trying to figure out how to do the OpenBSD port, I tried to
3940 clean up some of the code in src/runtime/. In particular, I dropped
3941 support for non-POSIX signal handling, added various comments,
3942 tweaked the code to reduce the number of compilation warnings, and
3943 renamed some files to increase consistency.
3944 * To support the new automatic configuration functionality in make.sh,
3945 the source file target-features.lisp-expr has been replaced with the
3946 source file base-target-features.lisp-expr and the machine-generated
3947 file local-target-features.lisp-expr.
3948 * fixed a stupid quoting error in make.sh so that using CMU CL
3949 "lisp -batch" as cross-compilation host works again
3951 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3953 * DESCRIBE no longer tries to call itself recursively to describe
3954 bound/fbound values, so that it no longer fails on symbols which are
3955 bound to themselves (like keywords, T, and NIL).
3956 * DESCRIBE now works on generic functions.
3957 * The printer now prints less-screwed-up representations of closures
3958 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3959 * A private symbol is used instead of the :EMPTY keyword previously
3960 used to mark empty slots in hash tables. Thus
3961 (DEFVAR *HT* (MAKE-HASH-TABLE))
3962 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3963 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3964 now does what ANSI says that it should. (You can still get
3965 similar noncompliant behavior if bang on the hash table
3966 implementation with all the symbols you get back from
3967 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3968 This breaks binary compatibility, since tests for equality to
3969 :EMPTY are wired into things like the macroexpansion of
3970 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3972 * There's now a minimal placeholder implementation for CL:STEP,
3973 as required by ANSI.
3974 * An obscure bug in the interaction of the normal compiler, the byte
3975 compiler, inlining, and structure predicates has been patched
3976 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3977 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3979 * Missing ordinary arguments in a macro call are now detected even
3980 when the macro lambda list contains &KEY or &REST.
3981 * The debugger no longer complains about encountering the top of the
3982 stack when you type "FRAME 0" to explicitly instruct it to go to
3983 the top of the stack. And it now prints the frame you request even
3984 if it's the current frame (instead of saying "You are here.").
3985 * As specified by ANSI, the system now always prints keywords
3986 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3988 * The default initial SIZE of HASH-TABLEs is now smaller.
3989 * Type information from CLOS class dispatch is now propagated
3990 into DEFMETHOD bodies, so that e.g.
3991 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3993 is now basically equivalent to
3994 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3995 (DECLARE (TYPE SINGLE-FLOAT X))
3997 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3998 operation, without having to do run-time type dispatch.
3999 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4000 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4001 * Since I don't seem to be making big file reorganizations very often
4002 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4003 is overflowing my ability to conveniently back them up), I've finally
4004 checked the system into CVS. (The CVS repository is on my home system,
4005 not at SourceForge -- putting it on SourceForge might come later.)
4006 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4007 high-level GC-NOTIFY-FOO functions send their output. (There's
4008 still very little control of where low-level verbose GC functions
4009 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4010 less than it used to -- the GC-NOTIFY-FOO functions are now under
4011 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4012 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4013 in only one place in the source code, and propagates it automatically
4014 everywhere that it's needed. Thus e.g. when I bump the version from
4015 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4016 * The C source files now include boilerplate legalese and documentation
4017 at the head of each file (just as the Lisp source files already did).
4018 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4019 to his page will be replaced with a link to his new CLiki service.
4021 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4023 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4024 * The build process now looks for GNU make under the default name "gmake",
4025 instead of "make" as it used to. If GNU make is not available as "gmake"
4026 on your system, you can change this default behavior by setting the
4027 GNUMAKE environment variable.
4028 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4029 can build without error under CMU CL.
4031 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4033 * There is now a partial SBCL user manual (with some new text and some
4034 text cribbed from the CMU CL manual).
4035 * The beginnings of a profiler have been added (starting with the
4036 CMU CL profiler and simplifying and cleaning up). Eventually the
4037 main interface should be through the TRACE macro, but for now,
4038 it's still accessed through vaguely CMU-CL-style functions and macros
4039 exported from the package SB-PROFILE.
4040 * Some problems left over from porting CMU CL to the new
4041 cross-compilation bootstrap process have been cleaned up:
4042 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4043 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4044 part when getting rid of the compiler *BACKEND* stuff.)
4045 ** The value of *NULL-TYPE* was screwed up, because it was
4046 being initialized before the type system knew the final
4047 definition of the 'NULL type. This screwed up several key
4048 optimizations in the compiler, causing inefficiency in all sorts
4049 of places. (I found it because I wanted to understand why
4050 GET-INTERNAL-RUN-TIME was consing.)
4051 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4052 PROCLAIM FTYPE information. Unfortunately this broke binary
4053 compatibility again, since now the forms output by DEFGENERIC
4054 to refer to functions which didn't exist in 0.6.3.
4055 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4056 can use the new (as of 0.6.3) transform for SOME into MAP into
4058 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4059 values to (INTEGER 0 1000000), so that the time code will no longer
4060 occasionally get blown up by Linux returning 1000000 microseconds
4061 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4062 its output conform to the ANSI spec. (Alas, this makes its output
4063 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4064 * A full call to MAP NIL with a single sequence argument no longer conses.
4065 * fixes to problems pointed out by Martin Atzmueller:
4066 * The manual page no longer talks about multiprocessing as though
4067 it were currently supported.
4068 * The ILISP support patches have been removed from the distribution,
4069 because as of version 5.10.1, ILISP now supports SBCL without us
4070 having to maintain patches.
4071 * added a modified version of Raymond Toy's recent CMU CL patch for
4072 EQUALP comparison of HASH-TABLE
4074 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4076 * The system still can't cross-compile itself with
4077 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4078 entails), but at least it can compile more of itself that way
4079 than it used to be able to, and various buglets which were uncovered
4080 by trying to cross-compile itself that way have now been fixed.
4081 * This release breaks binary compatibility again. This time
4082 at least I've incremented the FASL file format version to 2, so that the
4083 problem can be detected reliably instead of just causing weird errors.
4084 * various new style warnings:
4085 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4086 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4087 ** using the lexical binding of a variable named in the *FOO* style
4088 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4089 as specified by ANSI.
4090 * *RANDOM-STATE* is no longer automatically initialized from
4091 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4092 default behavior of the system is to repeat its behavior every time
4093 it's run. If you'd like to change this behavior, you can always
4094 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4095 old convention there was no comparably easy way to get the system to
4096 repeat its behavior every time it was run.
4097 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4098 FTYPE declarations has been removed, in favor of their ANSI
4099 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4100 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4101 the types of their sequence arguments can be determined at compile time.
4102 This is done through a new open code expansion for MAP which eliminates
4103 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4104 argument types can be determined at compile time.
4105 * The optimizer now transforms COERCE into an identity operation when it
4106 can prove that the coerced object is already of the correct type. (This
4107 can be a win for machine generated code, including the output of other
4108 optimization transforms, such as the MAP transform above.)
4109 * Credit information has been moved from source file headers into CREDITS.
4110 * Source file headers have been made more standard.
4111 * The CASE macro now compiles without complaining even when it has
4114 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4116 * (Note that the way that the PCL macroexpansions were rewritten
4117 to accommodate the change in DEFGENERIC below breaks binary
4118 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4119 not run under sbcl-0.6.2. Once we get out of alpha releases,
4120 i.e. hit release 1.0.0, we'll probably try to maintain binary
4121 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4122 and sbcl-1.4.4. Until then, however, it might be fairly common
4123 for maintenance releases to break binary compatibility.)
4124 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4126 * The reporting of unhandled signals has been changed to print some
4127 explanatory text as well as the report form. (Previously only
4128 the report form was printed.)
4129 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4130 it defines, so that the compiler no longer issues undefined function
4131 warnings for compiled-but-not-yet-loaded generic functions.
4132 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4133 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4134 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4135 section of the ANSI spec are supported.
4136 * The "" nickname for the "KEYWORD" package has been removed.
4137 The reader still handles symbol tokens which begin with a package marker
4138 as keywords, but it doesn't expose its mechanism for doing so in the
4139 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4140 * The system now issues STYLE-WARNINGs for contradictory TYPE
4141 proclamations. (Warnings for contradictory FTYPE proclamations would
4142 be nice too, but those can't be done usefully unless the type system
4143 is made smarter about FUNCTION types.)
4144 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4145 systematized, so that "*target-*.lisp is supposed to exist only on the
4146 target and imply that there's a related file which exists on the
4147 host, and *host-*.lisp is supposed to exist only on the host and imply
4148 that there's a related file which exists on the target. This involves a
4149 lot of renaming. Hopefully the acute confusion caused by the renaming
4150 will be justified by the reduction in chronic confusion..
4151 ** runtime-type.lisp -> early-target-type.lisp
4152 ** target-type.lisp -> late-target-type.lisp
4153 ** early-host-format.lisp -> early-format.lisp
4154 ** late-host-format.lisp -> late-format.lisp
4155 ** host-error.lisp -> misc-error.lisp
4156 ** early-error.lisp -> early-target-error.lisp
4157 ** late-error.lisp -> late-target-error.lisp
4158 ** host-defboot.lisp -> early-defboot.lisp
4159 ** code/misc.lisp -> code/target-misc.lisp
4160 ** code/host-misc.lisp -> code/misc.lisp
4161 ** code/numbers.lisp -> code/target-numbers.lisp
4162 ** code/early-numbers.lisp -> numbers.lisp
4163 ** early-host-type.lisp -> early-type.lisp
4164 ** late-host-type.lisp -> late-type.lisp
4165 ** host-typep.lisp -> typep.lisp
4166 ** load.lisp -> target-load.lisp
4167 ** host-load.lisp -> load.lisp
4168 ** host-disassem.lisp -> disassem.lisp
4169 ** host-insts.lisp -> insts.lisp
4170 ** byte-comp.lisp -> target-byte-comp.lisp
4171 ** host-byte-comp.lisp -> byte-comp.lisp
4172 ** host-signal.lisp -> signal.lisp
4173 ** host-defstruct.lisp -> defstruct.lisp
4174 ** late-target-type.lisp -> deftypes-for-target.lisp
4175 Furthermore, several other previously target-only files foo.lisp (e.g.
4176 hash-table.lisp and random.lisp) have been split into a target-and-host
4177 foo.lisp file and a target-only target-foo.lisp file, with their key type
4178 definitions in the target-and-host part, so that the cross-compiler will
4179 know more about target types.
4180 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4181 gone away. In their place are various *BACKEND-FOO* variables
4182 corresponding to the slots of the old structure.
4183 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4184 into the target SBCL has been fixed.
4185 * The chill.lisp system for loading cold code into a running SBCL
4187 * Support for the CMU CL "scavenger hook" extension has been removed.
4188 (It was undocumented and unused in the CMU CL sources that SBCL was
4189 derived from, and stale in sbcl-0.6.1.)
4190 * Various errors in the cross-compiler type system were detected
4191 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4192 (enabling various consistency checks). Many of them were fixed,
4193 but some hard problems remain, so the compiler is back to
4194 running without *TYPE-SYSTEM-INITIALIZED* for now.
4195 * As part of the cross-compiler type system cleanup, I implemented
4196 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4197 * I have started adding UNCROSS calls throughout the type system
4198 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4199 on cross-compiler input files will be able to go away, and various
4201 * CONSTANTP now returns true for quoted forms (as explicitly required
4204 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4206 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4207 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4208 (MAKE-LIST -1). In the longer term, it ought to become true
4209 that declarations are assertions even at SAFETY 1. For now, it's not
4210 quite true even at SAFETY 3, but it's at least more nearly true..
4211 (Note that this change seems to increases the size of the system by
4212 O(5%) and to decrease the speed of the compiler by 20% or more.)
4213 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4214 for the problem of printing dozens of lines of distracting information
4215 about low-level system machinery as part of the top stack frame
4216 on entry to the debugger when an undefined function was called.
4217 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4218 is not reset to COMMON-LISP-USER.
4219 * Compilation of stuff related to dyncount.lisp has been made conditional
4220 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4221 smaller. The various dyncount-related symbols have been moved into
4222 a new "SB-DYNCOUNT" package.
4223 * tty-inspect.lisp has been renamed to inspect.lisp.
4224 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4225 feature has gone away. (When we eventually port to other flavors of
4226 libc and/or Unix, we'll try to make the differences between flavors
4227 invisible at the user level.)
4228 * Various other *FEATURES* tags, and/or their associated conditionals,
4229 have been removed if obsolescent, or given better documentation, or
4230 sometimes given more-mnemonic names.
4232 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4234 * tidied up "make.sh" script
4235 * tidied up system directory structure
4236 * better "clean.sh" behavior
4237 * added doc/FOR-CMUCL-DEVELOPERS
4238 * many many small tweaks to output format, e.g. removing possibly-confusing
4239 trailing #\. character in DESCRIBE-INSTANCE
4240 * (EQUALP #\A 'A) no longer signals an error.
4241 * new hashing code, including EQUALP hashing
4242 * tidied up Lisp initialization and toplevel
4243 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4244 * command line argument processing
4245 * added POSIX-GETENV function to deal with Unix-ish environment variables
4246 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4247 terminating SBCL on EOF
4248 * non-verbose GC by default
4249 * There is no more "sbcl" shell script; the sbcl file is now the C
4250 runtime executable (just like CMU CL).
4251 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4253 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4254 symbol and package information as Lisp native symbol and package objects
4255 instead of strings naming symbols and strings naming packages. This way,
4256 whenever packages are renamed (as in warm init), debug information is
4257 transformed along with everything else.
4258 * tweaked the optimization policy declarations which control the building
4259 of SBCL itself. Now, among other things, the system no longer saves
4260 source location debugging information. (This helps two problems at once
4261 by reducing SBCL size and by keeping SBCL from trying to look for its
4262 sources -- which may not exist -- when reporting errors.)
4263 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4264 debugging and testing purposes
4265 * cleaned up printing, making the printer call PRINT-OBJECT for
4266 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4267 methods, giving nearly-ANSI behavior
4268 * converted almost all special variables to use *FOO* naming convention
4269 * deleted PARSE-TIME functionality, since it can be done portably
4270 * moved some files out of cold init into warm init
4271 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4273 * regularized formatting of source files
4274 * added an install.sh script
4275 * fixed ridiculous memory usage of cross-compiler by making
4276 compiler/alloc.lisp not try to do pooling unless it can hook
4277 itself into the GC of the cross-compilation host. Now the system
4278 builds nicely on my old laptop.
4279 * added :SB-ALLOC in target-features.lisp-expr
4280 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4281 was not implemented)
4282 * re-did condition handling and note reporting in the compiler. Notes
4283 are no longer handled by signalling conditions. Style warnings
4284 and warnings are handled more correctly and reported in such a way
4285 that it's easy to find one or the other in your output (so that you
4286 can e.g. figure out which of many problems caused COMPILE-FILE to
4288 * changed the severity of several compiler warnings from full WARNING
4289 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4290 compiler note reporting so that it doesn't use the condition system
4291 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4292 COMPILE-FILE command)
4293 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4294 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4295 no longer supported, since their implementation was deeply intertwingled
4296 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4297 * removed bogus "support" for compiler macros named (SETF FOO), and
4298 removed the compiler macro for SETF INFO (but only after making a fool
4299 of myself on the cmucl-imp mailing list by posting a bogus patch for
4300 DEFINE-COMPILER-MACRO..)
4301 * Compiled files containing forms which have side effects on the Lisp
4302 reader (such as DEFPACKAGE forms) are now handled more correctly.
4303 (Compiler queuing of top level lambdas has been suppressed by setting
4304 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4305 * deleted various currently-unused source files, e.g. gengc.lisp. They
4306 may be added back at some point e.g. when porting to other architectures,
4307 but until they are it's distracting to distribute them and to try to
4309 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4310 there were so many of them they're just distractions, and UNCROSS is
4311 known to be able to handle the current sources
4312 * moved PROFILE functionality into TRACE, so that it will be clear
4313 how the wrapping and unwrapping of functions when you profile them
4314 interacts with the wrapping and unwrapping of functions when you
4315 trace them. (Actually, the functionality isn't there yet, but at least
4316 the interface specification is there. Hopefully, the functionality will
4317 arrive with some future maintenance release.)
4318 * removed host-oops.lisp
4319 * changed signature of QUIT function to allow UNIX-CODE argument
4320 * fixed READ-SEQUENCE bug
4321 * tweaked verbose GC output so that it looks more like the progress
4322 output that ANSI specifies for functions like LOAD
4323 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4324 * added <http://sbcl.sourceforge.com> to the banner information printed by