1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
3 * minor incompatible change: the (unsupported) spinlock interface
4 has changed: free spinlock now has the value NIL, and a held spinlock
5 has the owning thread as its value.
6 * enhancement: name of a socket-stream is now "a socket" instead of
8 * bug fix: the cache used by the CLOS to store precomputed effective
9 methods, slot offsets, and constant return values is now thread and
11 * bug fix: generic function dispatch function updating is now thread
12 and interrupt safe (in the sense that the known issues have been
15 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
16 * new contrib: sb-cover, an experimental code coverage tool, is included
18 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
20 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
21 produce more efficient code on x86-64 if the compiler can determine
22 that (+ INDEX <constant>) does not require a bounds check and FOO
23 has an element type at least 8 bits wide.
24 * enhancement: a new, experimental synchronous timeout facility is
25 provided. Refer to SB-SYS:WITH-DEADLINE for details.
26 * enhancement: when a symbol name conflict error arises, the
27 conflicting symbols are always printed with a package prefix.
28 (thanks to Kevin Reid)
29 * enhancement: stepping is now once again supported on the SPARC and
30 MIPS platforms. (It is also now more likely to work on CheneyGC
32 * enhancement: sb-sprof can now also track and report accurate call
34 * bug fixes: the treatment of non-standard subclasses of
35 SB-MOP:SPECIALIZER is more correct.
36 * incompatible change: PURIFY no longer copies the data from the
37 dynamic space into the static and read-only spaces on platforms
38 that use the generational garbage collector
39 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
41 * bug fix: GC race condition occasionally resulting in crashes with
42 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
44 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
46 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
47 a STYLE-WARNING to be signalled (regression from 1.0.4.)
48 * bug fix: an asynchronous interrupt could previously leave the
49 system running with GC inhibited.
50 * bug fix: a DECLARE form evaluated at top-level now causes an error
51 rather than silently (or verbosely) returning NIL.
52 * bug fix: trying to dissassemble functions compiled at high DEBUG could
53 result in a error being signalled due to source form lookup errors.
54 (reported by Peter Graves)
56 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
57 * incompatible change: removed writer methods for host-ent-name,
58 host-ent-addresses -- changing the values did not update the DNS
60 * minor incompatible change: changed experimental JOIN-THREAD interface
61 * documentation: the manual now lists reader and writer methods
62 in class slot documentation sections. (thanks to Richard M Kreuter)
63 * documentation: unwinding from asyncronous events has been
65 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
66 in multithreaded application code.
67 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
68 platforms. (thanks to James Anderson for the optimization hint)
69 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
71 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
72 variants no longer cons.
73 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
74 their NOT- variants no longer cons.
75 * optimization: EQUAL hash tables no longer use SXHASH for objects
76 of all data types, but instead use an EQL hash for types for which
77 EQUAL is the same as EQL.
78 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
79 are significantly faster.
80 * optimization: new STRING-OUTPUT-STREAM implementation is much
81 faster for large outputs and conses 30% less on average.
82 * enhancement: XREF information is now collected for references made
83 to global variables using SYMBOL-VALUE and a constant argument.
84 * enhancement: SIGINT now causes a specific condition
85 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
86 * bug fix: FILE-POSITION used to signal an error in some cases where
87 ANSI requires it to return NIL.
88 * bug fix: ADJUST-ARRAY is now interrupt-safe.
89 * bug fix: adding and removing fd-handlers is now interrupt-safe.
90 * bug fix: inlined calls to C now ensure 16byte stack alignment on
92 * bug fix: bad type declaration in the CLOS implementation has
93 been fixed. (reported by James Anderson)
94 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
95 (thanks to Lutz Euler)
96 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
97 required by the CLHS. (reported by Samium Gromoff)
98 * bug fix: dead unbound variable references now signal an error.
99 (reported by Marco Monteiro)
100 * bug fix: / with an unused value was being deleted in safe code.
101 (thanks to Marco Monteiro and Kevin Reid)
102 * bug fix: number of characters that can be written onto a single
103 line in a file is unlimited.
104 * bug fix: some GC deadlocks caused by asynchronous interrupts have
105 been fixed by inhibiting interrupts for when GC is disbled.
106 * bug fix: some interrupt safety issues with GETHASH, (SETF
107 GETHASH), CLRHASH and REMHASH have been fixed.
108 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
109 specifier no longer causes infinite recursion.
110 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
111 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
112 * bug fix: an &environment argument with the correct variable information
113 is passed to macros that are expanded during byte compilation.
114 (reported by Samium Gromoff)
115 * bug fix: a logic error was causing the hash function for bignums to
116 have excessive amounts of collisions. (reported by Faré Rideau)
117 * bug fix: modifying the contents of an array could change the return
118 value of SXHASH on that array, which is only allowed for strings
119 and bit vectors. (bug introduced in 0.9.16)
120 * improvement: the x86-64/darwin port now passes all tests (except
121 for the debugger tests) but should still be considered
122 experimental until this is fixed.
123 * improvement: a style-warning is signaled for CASE (etc) clauses with
124 duplicate keys. (thanks to Kevin Reid)
125 * improvement: macos/x86 and macos/x86-64 now use mach exception
126 handlers for dealing with illegal instructions (for trapping and
127 error handling) and memory protection violations (for GC).
129 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
130 * new platform: experimental support for x86-64/darwin (MacOS).
131 * incompatible change: the thread-safe (on most platforms) getaddrinfo
132 and getnameinfo sockets functions are used instead of gethostbyaddr
133 and gethostbyname, on platforms where the newer functions are available.
134 As a result, the ALIASES field of HOST-ENT will always be NIL on these
136 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
137 don't print the startup banner, but behave as if --noinform was passed
138 as a command line argument. (thanks to Kevin Reid)
139 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
140 * optimization: code using alien values with undeclared types is much faster.
141 * optimization: the compiler is now able to open code SEARCH in more cases.
142 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
143 * bug fix: using standardized COMMON-LISP special variables as loop
144 variables no longer signals bogus package lock violations. (reported
146 * bug fix: declaring local loop variables to be of a range-limited type
147 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
148 (reported by Andras Simon)
149 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
150 bugs remain on x86-64.)
151 * bug fix: the #= and ## reader macros now interact reasonably with
152 funcallable instances.
153 * bug fix: type-checks for function arguments were compiled using the
154 compiler policy settings of the wrong lexical environment (in 1.0.2
156 * bug fix: SHADOW accepts characters as string designators, as required
157 by the spec (thanks to Eric Marsden)
158 * bug fix: fixed GC safety issues when foreign functions are called with
159 non-base strings as arguments
160 * bug fix: more consistent error output for fatal-compiler-warnings, like
162 * bug fix: functions with proper names were showing up as NIL in
163 backtraces (reported by Edi Weitz, regression in 1.0.2)
165 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
166 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
167 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
168 produce more efficient code on the x86 if the compiler can determine
169 that (+ INDEX <constant>) does not require a bounds check and FOO
170 has an element type at least 8 bits wide.
171 * bug fix: references to undefined variables are handled the same way
172 in toplevel forms as in the normal compiler
173 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
175 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
176 (thanks to Jon Buller)
177 * improvement: faster compilation times for complex functions
178 * improvement: added readlink support to SB-POSIX (thanks to Richard
181 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
182 * improvement: experimental support for mach exception handling on
183 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
185 * improvement: support for GBK external format.
186 (thanks to Chun Tian (binghe))
187 * improvement: the debugger now displays variables that have been closed
188 over, in code compiled with (DEBUG 2) or higher
189 * improvement: support for executable cores on NetBSD (thanks to
191 * new feature: added a RESTART-FRAME debugger command
192 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
193 be used to provide an implementation for FILE-POSITION on Gray streams
194 (thanks to Eric Marsden)
195 * new feature: add syslog, openlog and closelog support to SB-POSIX
196 (thanks to Richard Kreuter)
197 * optimization: the function call overhead in code compiled with
198 a high DEBUG optimization setting is significantly
199 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
200 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
201 (thanks to Tony Martinez)
202 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
203 evaluated once (reported by Kevin Reid)
204 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
205 works again (thanks to Kilian Sprotte)
206 * bug fix: an error is signaled for tagbodies with duplicate tags
207 (thanks to Stephen Wilson)
208 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
209 * bug fix: Win32 port can now handle foreign code unwinding Lisp
210 stack frames from alien callbacks.
211 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
212 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
213 65k pages (thanks to David Woodhouse)
214 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
216 changes in sbcl-1.0.1 relative to sbcl-1.0:
217 * new platform: FreeBSD/x86-64, including support for threading.
218 * new feature: the compiler stores cross-referencing information
219 abount function calls (who-calls), macroexpansion (who-macroexpands)
220 and special variables (who-binds, who-sets, who-references) for code
221 compiled with (< SPACE 3). This information is available through the
222 sb-introspect contrib.
223 * new feature: users may subclass SEQUENCE, and have instances of
224 these classes interoperate with standard Common Lisp functions if
225 a number of methods are defined. (This feature is experimental
226 and the interface subject to change based on feedback from SBCL
227 users and the general community)
228 * improvement: sb-sprof traces call stacks to an arbitrary depth on
229 x86/x86-64, rather than the previous fixed depth of 8
230 * improvement: another pthread back-end of mutex "pthread-futex".
231 * bug fix: non-ascii command-line arguments are processed correctly
232 (thanks to Yaroslav Kavenchuk)
233 * bug fix: non-required arguments were not passed correctly when a method
234 defined using DEFMETHOD was called from a mop-generated method using
235 CALL-NEXT-METHOD (reported by Pascal Costanza)
236 * bug fix: recursion is now permitted in accessors through
237 SLOT-UNBOUND. (reported by Pascal Costanza)
238 * bug fix: an error was signaled at startup if the HOME environment
239 variable was defined, but had an empty value (reported by Peter Van Eynde)
240 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
241 signal a PROGRAM-ERROR, not a BUG.
242 * bug fix: SB-POSIX:READDIR works when built with large file support.
243 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
245 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
246 for better automatic indentation support. (thanks to Matt Pillsbury)
247 * optimization: loading generic functions no longer takes O(n^2) time,
248 proportional to the amount of methods in the generic function
249 (reported by Todd Sabin and Jeremy Brown)
250 * optimization: the FIND and POSITION family of sequence functions
251 are significantly faster on arrays whose element types have been
253 * improvements to the Windows port:
254 ** Intermittent heap corruption problems have been fixed. (thanks
255 to Alastair Bridgewater)
256 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
258 ** Lisp is able to unwind foreign exception frames from alien
259 callbacks. (thanks to Alastair Bridgewater)
261 changes in sbcl-1.0 relative to sbcl-0.9.18:
262 * improvement: experimental support for threading on FreeBSD/x86.
263 (thanks to NIIMI Satoshi)
264 * improvement: runtime option --dynamic-space-size can be used
265 to set the size of the dynamic space reserved on startup.
266 * improvement: floating point modes in effect are now saved in
267 core, and restored on startup.
268 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
269 startup, not time since first call to GET-INTERNAL-REAL-TIME.
270 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
271 threads are not running after *SAVE-HOOKS* have run.
272 * improvement: writes to CLOS instance slots are type-checked in code
273 compiled with (SAFETY 3)
274 * improvement: floating-point exception handling on FreeBSD (thanks to
276 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
277 (thanks to Zach Beane)
278 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
280 * improvement: added support for the Shift-JIS external format.
281 (contributed by NIIMI Satoshi)
282 * improvement: callbacks are supported on Linux/PPC. (thanks to
284 * bug fix: compiler bug triggered by a (non-standard) VALUES
285 declaration in a LET* was fixed. (reported by Kaersten Poeck)
286 * bug fix: file compiler no longer confuses validated and already
287 dumped structurres. (reported by Kaersten Poeck)
288 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
289 fill-pointer signals a type-error as required. (thanks to
291 * bug fix: disassemly of funcallable instances works.
292 * bug fix: single stepping on PPC.
293 * bug fix: fix thread-safety problems in the type system (generally
294 manifesting as nonsensical errors like "STRING is a bad type specifier
295 for sequences" or "The value 1 is not of type FIXNUM")
296 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
297 (reported by Josip Gracin)
298 * bug fix: an error is signaled for attempts to displace arrays with
299 incompatible element types (thanks to Mario Mommer)
300 * bug fix: more correct handling of wide characters in debug info
301 (bug reported by Attila Lendvai and fixed by Juho Snellman)
302 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
303 and don't cause extra consing
304 * optimization: MAP and MAP-INTO are significantly faster on vectors
305 whose elements types have been declared.
306 * Improvements to SB-SPROF:
307 ** Support for allocation profiling
308 ** Reduced profiling overhead, especially for long profiling runs
309 * Improvements to the Windows port:
310 ** floating point exceptions are now reported correctly.
311 ** stack exhaustion detection works partially.
312 ** more accurate GET-INTERNAL-REAL-TIME.
313 ** better RUN-PROGRAM behaviour in case of errors in the child
315 ** PROBE-FILE now simplifies pathnames correctly.
316 ** DIRECTORY now works correctly with :WILD-INFERIORS.
317 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
318 1st 1970 (although time timezone data will be unreliable).
319 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
321 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
322 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
323 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
325 * bug fix: two potential GC deadlocks affecting threaded builds.
326 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
327 returns T (reported by Anton Kazennikov)
328 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
329 with non-variable places
330 * bug fix: remove a race condition in the setting of
331 funcallable-instance functions, this should make threaded CLOS
332 code more stable against memory faults.
333 * bug fix: corruption of specials when unbinding is interrupted by an
334 asynchronous unwind (reported by Hannu Koivisto)
335 * improvement: the debugger will now also display local variables that
336 are only used once, for code compiled with a DEBUG optimization quality
339 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
340 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
341 * incompatible change: External-format support for FFI calls. The
342 SB-ALIEN:C-STRING no longer implies an ASCII
343 external-format. Instead the string is subject to external-format
344 conversion. Additionally return values of type SB-ALIEN:C-STRING
345 are by default of type (SIMPLE-ARRAY CHARACTER), not
346 SIMPLE-BASE-STRING. For an alien type that behaves like the old
347 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
348 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
349 * incompatible change: SB-EXT package no longer contains the
350 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
351 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
353 * incompatible change: the single-stepper is no longer available
354 on Alpha, Mips and Sparc platforms.
355 * minor incompatible change: the direct superclasses of
356 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
357 not (STANDARD-OBJECT FUNCTION). This makes the
358 class-precedence-lists of GENERIC-FUNCTION and
359 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
361 * minor incompatible change: the default stream external format on
362 non-unicode SBCL is no longer detected from the locale, but is always
364 * new feature: Add a version of evaluator that uses an interpreter instead
365 of the compiler. EVAL still uses the compiler by default, to switch it
366 to use the interpreter, set the value of the variable
367 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
368 * minor incompatible change: the single-stepper REPL has been merged
369 with the normal debugger (see the "Stepping" heading of the debugger help
370 for more details). The debugger command STEP will no longer switch
371 to the single-stepper REPL.
372 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
373 for a type now works.
374 * bug fix: loading of default sysinit file works. (thanks to Leonid
376 * bug fix: better detection of circularities in the file-compiler.
377 (reported by Marco Monteiro)
378 * bug fix: the CL pathname functions now work with files that have
379 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
380 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
381 systems with Unix98 pty semantics.
382 * bug fix: ASDF-INSTALL will now work with bsd tar.
383 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
385 * bug fix: timers expiring in dead threads no longer cause a
386 type-error (reported by Paul "Nonny Mouse").
387 * bug fix: thanks to more lightweight single-stepper instrumentation,
388 code compiled with (DEBUG 3) will compile and execute significantly faster,
389 and will have more accurate type-inferencing than before
390 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
391 whose bindings are modified
392 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
393 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
394 ** replace dummy implementations of CL:MACHINE-INSTANCE and
395 CL:SOFTWARE-VERSION with proper version
397 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
398 * feature: implemented the READER-METHOD-CLASS and
399 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
400 as specified by AMOP.
401 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
403 * optimization: faster LOGCOUNT implementation on x86 and x86-64
404 (thanks to Lutz Euler)
405 * optimization: hashing of general arrays and vectors has been
406 improved. (reported by Any Fingerhut)
407 * enhancement: SB-INTROSPECT is now able to find definitions of
408 profiled functions. (thanks to Troels Henriksen)
409 * enhancement: compiler-macro expansion applies now to FUNCALL forms
410 as well, allowing compiler-macros for SETF-functions to expand.
411 * enhancement: step-instrumentation no longer wraps known
412 single-value functions with multiple-value context, allowing
413 better type inference.
414 * fixed bug #337: use of MAKE-METHOD in method combination now works
415 even in the presence of user-defined method classes. (reported by
416 Bruno Haible and Pascal Costanza)
417 * fixed bug #339(c): if there are applicable methods not part of any
418 long-form method-combination group, call INVALID-METHOD-ERROR.
419 (reported by Bruno Haible)
420 * fixed bug #361: the :FUNCTION initarg in the protocol for
421 initialization of methods can now be used to override
422 internally-produced optimized functions. (reported by Bruno
424 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
425 system-provided lambda expression no longer cause warnings about
426 unbound #:|pv-table| symbols.
427 * bug fix: improved the handling of type declarations and the
428 detection of violations for keyword arguments with non-constant
430 * bug fix: potentially erronous calls to PATHNAME and
431 MERGE-PATHNAMES were being flushed in some cases.
432 (reported by Richard Kreuter)
433 * bug fix: compiled calls to TYPEP were mishandling obsolete
434 instances. (reported by James Bielman and Attila Lendvai)
435 * bug fix: format strings with ~> without matching ~< no longer
436 trigger an AVER, but signal an understandable error instead.
437 (reported by Antonio Martinez)
438 * bug fix: specifying an output-file without a pathname-type for
439 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
440 of input-file instead of "fasl". (reported by Robert Dodier)
441 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
442 (reported by James Y Knight).
443 * bug fix: compiler-macros-function did not consider the environment
444 argument for shadowing by local functions.
445 * bug fix: compiler-macros expansion was inhibited by local INLINE
447 * bug fix: inline expansions of known functions were subject to
448 step-instrumentation in high debug policies, leading to problems
450 * bug fix: compiler failed to differentiate between different CONS
452 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
453 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
454 * bug fix: #368: incorrect use of expressed vs. upgraded array
456 * bug fix: #306a: more precise unions of array types.
457 * thread-safety improvements:
458 ** CONDITION-WAIT could return early on Linux, if the thread was
459 interrupted and subsequently continued with SIGCONT.
460 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
462 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
463 * added support for the ucs-2 external format. (contributed by Ivan
465 * minor incompatible change: pretty printing of objects of type
466 (cons symbol) is, in the default pprint-dispatch-table, now
467 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
469 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
470 called later in a class's lifetime, possibly as late as when the
471 first instance of the class is created. Previously,
472 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
473 class became finalizeable.
474 * fixed bug: FILE-POSITION sometimes returned inconsistent results
475 for multibyte external-format streams. (thanks to "vbzoli")
476 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
477 with :ALLOCATION :CLASS inherited from superclasses of the
479 * fixed bug: anonymous classes can now be created using the :NAME
480 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
481 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
482 * fixed bug: core-files saved with :EXECUTABLE T can again be
483 executed when SBCL_HOME isn't set. (reported by James Knight)
484 * fixed bug: toplevel LOCALLY forms with declarations could
485 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
486 * fixed bug: printing from several different threads using different
487 values of *print-case* could cause invalid output, due to
488 some internal special variables of the printer not being bound
489 thread-locally (reported by Max Mikhanosha)
490 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
491 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
492 (reported by Pascal Costanza)
493 * fixed bug: FINALIZE-INHERITANCE is called from
494 REINITIALIZE-INSTANCE on classes when the class has previously
495 been finalized, as required by AMOP.
496 * minor code generation optimizations:
497 ** better register allocation in CLOS dispatching functions
498 ** overflow detection when coercing signed bytes to fixnums on x86-64
499 ** is now implemented with one IMUL instruction instead of three shifts
500 ** more efficient bit-vector access on x86 and x86-64
501 ** more efficient access to raw structure slots on x86 and x86-64
502 * fixed some bugs revealed by Paul Dietz' test suite:
503 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
506 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
507 * feature: thread support on Solaris/x86, and experimental thread support
509 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
511 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
512 default method for SLOT-UNBOUND.
513 * minor incompatible change: prevent the user from specializing the
514 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
515 somewhat counter to the intent of the protocol, I (CSR) think, and
516 additionally it just doesn't work in SBCL as currently
517 implemented, thanks to optimizations (that are always valid for
518 the other three SLOT-VALUEish functions, but not for the setter).
519 * minor incompatible change: the :SB-LDB feature is now enabled by
520 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
521 the low-level debugger.
522 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
523 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
524 down to an SB-SYS:FD-STREAM, allowing direct communication in
525 more cases. Temporary files and pipes are still used as a fallback
527 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
528 on the non-reentrant C functions localtime(3) and gmtime(3).
529 * bug fix: global optimization policy was not visible in LOCALLY and
531 * bug fix: class objects can be used as specializers in methods.
532 (reported by Pascal Costanza)
533 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
534 with a type error. (reported by blitz_ on #lisp)
535 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
536 failed with a type error. (reported by Pascal Bourguignon)
537 * bug fix: merging pathnames against defaults with :DIRECTORY
538 starting with '(:RELATIVE :BACK) should preserve the :BACK.
539 (reported by James Y Knight)
540 * bug fix: saving large (>2GB) cores on x86-64 now works
541 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
542 constant index and a (simple-array (signed-byte 32)) array
543 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
544 error (patch by Robert J. Macomber)
545 * feature: TIME now displays the amount of run-time spent in GC
546 * bug fix: The debugger now does a better job of respecting
547 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
548 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
549 workaround for bug 403.)
550 * bug fix: floating point exception handling now works on FreeBSD
551 (thanks to NIIMI Satoshi)
552 * fixed some bugs revealed by Paul Dietz' test suite:
553 ** MISC.641: LET-conversion were not supposed to work in late
555 * improvements to the Win32/x86 port:
556 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
557 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
559 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
560 * new feature: source path information is generated for macro-expansion
561 errors for use in IDE's like Slime (thanks to Helmut Eller)
562 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
563 * bug fix: compiler does not loop forever on an invalid type in
565 * improvement: compilation of most CLOS applications is significantly
567 * optimization: added a limited bytecode compiler for simple toplevel
568 forms, speeding up compilation and FASL loading
569 * bug fix: the statistical profiler now properly distinguishes anonymous
572 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
573 * minor incompatible change: in sbcl-0.9.11 (but not earlier
574 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
575 into subclasses of the finalized class. Now user calls to
576 FINALIZE-INHERITANCE finalize just the one class, and calls by the
577 system return before any subclasses are finalized.
578 * minor incompatible change: The reader no longer ignores errors
579 regarding non-existent packages in #+ and #- feature tests.
580 * new feature: command line options --no-sysinit, --no-userinit to
581 inhibit loading the corresponding init files
582 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
583 for sending data through UDP sockets (thanks to François-René Rideau)
584 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
585 error is signalled instead (thanks to François-René Rideau)
586 * minor incompatible change: Error signalling behaviour of lexical
587 operations violating package locks has changed slightly. Refer to
588 documentation on package locks for details.
589 * enhancement: EVAL can process IF-expressions without resorting to the
591 * bug fix: Some operations involving SIN, COS, or TAN weren't being
592 constant-folded properly due to missing out-of-line functions.
593 (reported by Mika Pihlajamäki)
594 * bug fix: LISTEN sometimes returned T even in cases where no data was
595 immediately available from the stream
596 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
597 were reversed. (reported by Levente Mészáros)
598 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
599 when given a vector displaced to an adjustable array. (reported
601 * bug fix: derivation of float boundaries from numbers outside the
602 appropriate float range (reported by John Wiseman)
603 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
605 * bug fix: precomputing cache entries for generic functions with
606 some subclasses of specializers as yet invalid does not attempt to
607 fill a cache line with a negative offset. (reported by Levente
609 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
610 allow condition class objects as arguments to DOCUMENTATION and
611 (SETF DOCUMENTATION); only find and set documentation for
612 structure names for the STRUCTURE doc-type. (suggested by Gary
614 * improvements to the Win32/x86 port:
615 ** (user-homedir-pathname) and default initialization file
616 locations now know about the user's "Documents and Settings"
617 directory (thanks to Yaroslav Kavenchuk)
618 ** run-program is implemented (thanks to Mike Thomas)
619 ** sockets support (thanks to Timothy Ritchey)
620 ** better backtrace support (thanks to Alastair Bridgewater)
621 ** sb-grovel supported
622 ** asdf-install and sb-posix work somewhat
623 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
624 * improvements to the Solaris/x86 port:
625 ** works on Solaris 11/Solaris Express
626 ** floating-point exception handling support
627 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
628 * fixed some bugs revealed by Paul Dietz' test suite:
629 ** REMOVE-METHOD returns its generic function argument even when
630 the method is not one of the generic functions' methods.
631 ** objects declared as MEMBER types can be admissible arguments to
633 ** printing characters should simply be printed by the FORMAT ~:C
635 ** compiler failure when compiling functions with hairy constant
636 defaults for optional parameters.
637 ** compiler produces wrong code when MAYBE-INLINE-expanding a
638 function, which is already optimized.
640 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
641 * new platform: experimental support for SBCL x86/Darwin, including
642 MacOS X 10.4.5 on Intel.
643 * new feature: Unicode character names are now known to the system
644 (through CHAR-NAME and NAME-CHAR).
645 * new feature: the filesystem location of SBCL's core file is
646 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
647 * minor incompatible change: the contrib modules SB-POSIX and
648 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
649 this change is to make it easier to distribute
650 location-independent binaries.
651 * bug fix: as implied by AMOP, standardized classes no longer have
652 slots named by external symbols of public packages. (reported by
654 * optimization: calling functions via constant symbols -- as in
655 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
656 via the function object as in (FUNCALL #'FOO).
657 * optimization: on x86 and x86-64, the calling convention for the
658 case of dealing with an unknown number of values has been altered
659 to be friendlier to the prediction heuristics implemented,
660 particularly with reference to CALL and RET pairing. (thanks to
661 Alastair Bridgewater)
662 * optimization: on x86, the code for access to thread-local symbol
663 values and binding/unbinding of thread-local symbols is smaller.
664 (thanks to Alastair Bridgewater)
665 * enhancement: CONSTANTP is now able to determine constantness of
666 more complex forms, including calls to constant-foldable standardized
667 functions and some special forms beyond QUOTE.
668 * fixed bug: occasional GC crashes on Solaris/x86
669 * optimization: x86-64 supports stack allocation of results of simple
670 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
671 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
672 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
673 (thanks to James Knight)
674 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
675 by the standard (thanks to Stephen Compall)
677 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
678 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
679 be used for bundling the runtime and the core file into one
680 executable binary. This feature is not currently supported on all SBCL
681 platforms. (thanks to James Bielman and NIIMI Satoshi)
682 * new feature: a generational or ephemeral garbage collector is now
683 the default on the PowerPC platform (both Linux and Darwin). The
684 old Cheney (stop and copy) collector is a build-time option.
685 (thanks to Cyrus Harmon, after Raymond Toy)
686 * minor incompatible change: the method by which SBCL finds its
687 contributed modules has changed; it no longer relies on symbolic
688 links from an $SBCL_HOME/systems directory, but searches directly
689 in the subdirectories of $SBCL_HOME.
690 * enhancement: the dynamic heap size on the Linux/PPC platform is
691 markedly larger, even using the older Cheney garbage collector.
692 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
693 certain complicated string types. (reported by Gary King)
694 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
695 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
696 * fixed bug: DOCUMENTATION on structures no longer signals an
697 error if no structure type of the right name exists. (reported by
699 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
700 applicable non-standard primary or :AROUND method on
701 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
702 no longer cause unbound variable errors. (reported by Kevin Reid)
703 * fixed bug: metacircle resolution in cases where methods have slots
704 added before the slots from STANDARD-METHOD. (reported by Jean
706 * fixed bug: the Power PC assembly code for calling into lisp
707 sometimes computed its return address wrongly (depending on the
708 15th bit of the address, fixed at link time). This used to
709 manifest itself in a segmentation violation while building PCL.
710 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
711 many others over the years)
712 * fixed bug: ROOM no longer reports silly numbers for stack usage if
713 the stack pointer (treated as a signed integer) is negative.
714 (thanks to Peter van Eynde)
716 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
717 * new platform: experimental support for the Windows operating
718 system has been added. (thanks to Alastair Bridgewater)
719 * fixed several bugs in and robustified the PPC FFI (including
720 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
721 * bug fix: allow non-simple string symbol names (reported by Paul
723 * bug fix: interrupt handling on NetBSD (thanks to Richard M
725 * bug fix: saving a core corrupted callbacks on x86/x86-64
726 * bug fix: closed a loophole in metacircularity detection and
727 grounding in the PCL implementation of CLOS.
728 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
729 (reported by tomppa on #lisp)
730 * bug fix: clear the direction flag on Lisp -> C transitions, as
731 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
732 (reported by Andrej Grozin and Hendrik Maryns)
733 * optimization: major improvements to GC efficiency on GENCGC platforms
734 * optimization: faster implementation of EQUAL
735 * optimization: emit more efficient opcodes for some common
736 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
737 * optimization: several other minor code-generation improvements on x86-64
738 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
740 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
741 * minor incompatible change: (SETF CLASS-NAME) and (SETF
742 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
743 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
744 by ANSI as a generic function, and for consistency (SETF
745 GENERIC-FUNCTION-NAME) is treated likewise.
746 * fixed bug #233.b: make constraint propagation notice when a variable
747 value is changed after it is referenced but before it is used
748 * fixed bug #296: no more arbitrary behaviour from filenames with
749 odd characters as --load arguments. (reported by Adam Warner)
750 * fixed bug #390: :CHARACTER-SET pathname components now work as
751 expected. (reported by Tim Daly Jr)
752 * fixed bug #391: complicated :TYPE intersections in slot
753 definitions no longer cause an error in PCL internals.
754 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
755 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
756 returning the number of octets which would be written to the
757 file-stream. (thanks to Robert J. Macomber)
758 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
759 arguments to the internal stream functions. (thanks to David
761 * bug fix: stack exhaustion is now handled on the x86/SunOS
762 platform. (thanks to Daisuke Homma)
763 * bug fix: more accurate ROOM results on GENCGC platforms
764 * optimization: improved type inference for arithmetic-for
765 index variables in LOOP
766 * optimization: faster floating-point SQRT on x86-64
767 * fixed some bugs revealed by Paul Dietz' test suite:
768 ** DOCUMENTATION returns NIL instead of "" for method combinations
769 that don't have a docstring
771 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
772 * minor incompatible change: (SETF CLASS-NAME) and (SETF
773 GENERIC-FUNCTION-NAME) are no longer generic functions, and
774 therefore are not customizeable by user code (as seems to be at
775 least permitted and maybe required by AMOP). As a consolation,
776 however, the SBCL implementation of these functions now calls
777 REINITIALIZE-INSTANCE as specified by AMOP.
778 * bug fix: slot-definition documentation is propagated as per ANSI
779 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
780 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
781 argument, as do eslotd initialization methods. (from Pascal
782 Costanza's "Closer" project)
783 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
784 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
786 * bug fix: it is now possible to have more than one subclass of
787 STANDARD-GENERIC-FUNCTION without causing stack overflow.
788 (reported by Bruno Haible, Pascal Costanza and others)
789 * bug fix: the dependent update protocol now works for generic
790 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
792 * bug fix: condition-class instances corresponding to
793 DEFINE-CONDITION forms are now created eagerly. (reported by
794 Kalle Olavi Niemitalo on comp.lang.lisp)
795 * bug fix: floating point printing is more accurate in some
796 circumstances. (thanks to Simon Alexander)
797 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
798 merged with *DEFAULT-PATHNAME-DEFAULTS*.
799 * bug fix: callbacks on OS X now preserve stack-alignment.
800 (thanks to Cyrus Harmon)
801 * enhancement: the x86-64 disassembler is much better at
802 disassembling SSE instructions. (thanks to Lutz Euler)
803 * enhancement: CHECK-TYPE on a variable now allows type-inference
804 (thanks to Vincent Arkesteijn)
805 * optimization: improved performance of EUC-JP external format.
806 (thanks to NIIMI Satoshi)
807 * optimization: performance improvements to IO on file streams of
808 :ELEMENT-TYPE CHARACTER
809 * optimization: much faster memory allocation on x86-64
810 * optimization: faster garbage collections (latency and throughput)
812 * optimization: faster bignum subtraction and division on x86 and x86-64
813 (thanks to Lutz Euler)
814 * optimization: more accurate type-inference for loops that use a
815 floating point index variable or a negative step.
817 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
818 * bug fix: add a workaround to SBCL looping infinitely at startup on
819 Linux kernels with apparently buggy implementations of personality().
820 (thanks to Svein Ove Aas)
821 * bug fix: Unicode symbols are correctly printed in LDB backtraces
822 (thanks to David Lichteblau)
823 * bug fix: local bindings shadow global symbol macros for
824 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
825 * optimization: non-open coded uses of numeric comparison operators
826 (e.g. >) no longer cons when called with more than one parameter
827 on platforms supporting dynamic-extent allocation.
828 * enhancement: saving cores with foreign code loaded is now supported
829 on MIPS/Linux in addition to the previously supported platforms.
830 * bug fix: FIND-RESTART now tests for activity, not applicability when given
831 a restart object as identifier. (reported by Helmut Eller for CMUCL)
832 * bug fix: division by zero in sb-sprof when no samples were collected
833 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
834 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
835 (thanks to Alexander Kjeldaas)
836 * new feature: ignoring the return values of destructive functions
837 like NREVERSE signals a compile-time style-warning.
838 (thanks to Kevin Reid)
839 * new feature: the alignment of alien structure fields can be
840 explicitly specified. (thanks to Cyrus Harmon)
841 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
842 and dump core on SIGQUIT
844 ** incompatible change: threads do not inherit values of specials
845 from their parents (see manual)
846 ** bug fix: threads stacks belonging to dead threads are freed by the
847 next exiting thread, no need to gc to collect thread stacks anymore
848 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
849 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
850 ** bug fix: dynamic variable and thread start related gc lossage
851 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
853 * fixed some bugs revealed by Paul Dietz' test suite:
854 ** SUBTYPEP is slightly more accurate on heinously complicated
855 CONS types where some of the members have uncertain (in the
856 NIL, NIL sense) type relationships to each other.
858 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
859 no longer purified unless :PURIFY T is explicitly specified.
860 ** Non-purified cores are significantly smaller than before
862 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
863 * new feature: timers based on Zach Beane's excellent timer package
864 * added support for the following external formats: koi8-u,
865 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
866 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
867 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
868 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
869 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
870 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
871 * incompatible change: a threaded SBCL will no longer revert to
872 non-threaded mode on non-NPTL systems, but refuse to start entirely.
873 * bug fix: interrupts are disabled until startup is complete; no
874 more sigsegvs when receiving a signal to soon
875 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
877 * bug fix: add a workaround for the memory randomization features in
878 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
879 workaround will only be in effect on systems with the proc filesystem
881 * bug fix: printing objects of type HASH-TABLE signals a
882 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
884 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
885 have been running for over 50 days. (reported by Gilbert Baumann)
886 * bug fix: the logic for getting names of functions gets less
887 confused when confronded with alternate-metaclass
888 funcallable-instances. (reported by Cyrus Harmon)
889 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
890 consistent, even on internal alternate-metaclass objects.
891 * bug fix: flush closure information collected by physical
892 environment analysis prepass before the main pass. (bug reported
894 * bug fix: compiler pack phase does not modify a hashtable, which is
895 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
896 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
897 may be expanded. (reported by Alexander <malishev> on c.l.l)
898 * bug fix: no more style-warnings when compiling code using LAST
899 under high-SPEED low-SPACE settings. (reported by David Wallin
901 * fixed some incompatibilities between SBCL's MOP and the MOP
903 ** the METAOBJECT class is now implemented;
904 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
905 STANDARD-OBJECT, as required;
906 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
907 now compatible (as required by VALIDATE-SUPERCLASS); there
908 remains a consistency requirement over the presence or absence
909 of the FUNCTION class in the superclass list of finalized
910 classes; see the manual for more details;
911 ** the :AROUND method for COMPUTE-SLOTS on
912 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
913 requested slot ordering.
915 ** bug fix: parent thread now can be gc'ed even with a live
917 ** bug fix: binding a special with PROGV to no value is not
918 the same as that symbol not having been bound (thanks to
920 ** bug fix: binding specials is thread safe (thanks to
922 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
923 and SB-SPROF do not die with 'no handler for signal XX in
924 interrupt_handle_now(..)' anymore
925 ** bug fix: WITH-TIMEOUT works with multiple running threads
926 * fixed some bugs revealed by Paul Dietz' test suite:
927 ** ENSURE-GENERIC-FUNCTION should take a method class object for
928 the :method-class keyword argument.
930 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
931 * new port: the Solaris operating system on x86 processors is now
932 mostly supported, though some rough edges in the environment
933 remain. (thanks to Daisuke Homma)
934 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
935 space for its heap. (thanks to Thiemo Seufer)
936 * optimization: SBCL on MIPS platforms now supports dynamic-extent
937 closures. (thanks to Thiemo Seufer)
938 * minor incompatible change: eof selects abort in the debugger.
939 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
940 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
942 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
943 deprecated, and will go away in a future revision of SBCL.
944 * minor incompatible change: GC-ON and GC-OFF are no longer
945 implemented with a counter, it does not matter how many times gc
946 is switched on or off
947 * bug fix: discriminating functions for generic function classes
948 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
949 make invalid assumptions about method precedence order. (reported
951 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
952 platforms. (thanks to Thiemo Seufer)
953 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
954 * bug fix: correct pathname printing with printer escaping is on.
955 (thanks to Kevin Reid)
956 * bug fix: complex VOP definitions in "user-space" no longer trigger
957 package locks. (reported by Zach Beane)
958 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
959 for funcallable-instance functions, and is no different from
961 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
962 characters in the current readtable. (reported by Nicholas Neuss)
963 * bug fix: SUBTYPEP on various CONS types returns more a more
964 accurate acknowledgment of its certainty.
965 * optimizations: REMOVE-DUPLICATES now runs in linear time on
966 lists in some cases. This partially fixes bug 384.
967 * flush all standard streams before prompting in the REPL and the
969 * bug fix: signal handling and triggering gc do not conflict
970 directly anymore, in particular a high frequency sb-sprof does
971 not prevent gc from running
972 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
973 approximation for timezone and DST information between the
974 universal time epoch and the smallest negative 32-bit time_t.
975 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
976 year 1899 so long as the encoded time is non-negative.
977 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
978 the correct answer for non-constant rotate amounts when performing
979 an inline 32-bit rotation.
981 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
982 there is only one thread in the session
983 ** bug fix: memory leak for streams created in one thread and
984 written to in another
985 ** bug fix: lockup when compiled with gcc4
986 ** bug fix: race that allows the gc to be triggered when gc is
988 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
989 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
991 ** bug fix: finished threads are gc'ed properly
992 * fixed some bugs revealed by Paul Dietz' test suite:
993 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
994 has arguments with a different set of applicable methods from
995 the orignal arguments.
996 ** The type error thrown by MAP now has a correct expected-type
998 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
999 name a compiled function.
1000 ** Three MISC tests where a large bignum was improperly coerced to
1001 a float that couldn't represent that bignum during type
1002 derivation were fixed.
1003 ** SUBTYPEP can now handle types involving the negation of a
1004 list-form FUNCTION type.
1005 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1006 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1007 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1009 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1010 * New feature: Experimental support for bivalent streams: streams
1011 opened with :element-type :default now allow character and binary
1012 (unsigned-byte 8) I/O
1013 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1014 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1016 * bug fix: no more highly sporadic "couldn't check whether ~S is
1017 readable" when reading a stream and an interrupt hits in the middle
1018 of a select system call
1019 * compiler better recognizes complex arrays (reported by Eduardo
1021 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1022 for variables on SBCL built with linkage-tables. (reported by Luis
1024 * various error reporting improvements.
1025 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1026 (Thanks to Thiemo Seufer)
1027 * bug fix: nested reader invokations work correctly
1028 * bug fix: it is possible to have simultaneous references to foreign
1029 code and foreign data with the same name.
1031 ** added x86-64 support
1032 ** incompatible change: the threading api now works with thread
1033 objects instead of thread ids
1034 ** bug fix: threads are protected from signals and interruption when
1035 starting up or going down
1036 ** bug fix: a race where an exiting thread could lose its stack to gc
1037 ** bug fix: don't halt on infinite error in threads if possible
1038 ** fixed numerous gc deadlocks introduced in the pthread merge
1039 ** bug fix: fixed thread safety issues in read and print
1040 ** bug fix: debugger doesn't hang on session lock if interrupted at
1041 an inappropriate moment
1042 ** bug fix: run-program is now thread safe(r)
1043 ** bug fix: inner with-recursive-lock no longer releases the mutex
1044 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1045 * fixed some bugs revealed by Paul Dietz' test suite:
1046 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1047 EXPECTED-TYPE slots filled.
1048 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1050 ** MAKE-CONDITION accepts classes as type-designators.
1051 ** COMPILE may never return NIL.
1052 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1053 range before calling Unix time functions
1055 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1056 * numerous signal handling fixes to increase stability
1057 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1058 * minor incompatible change: we now correctly canonize default
1059 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1060 as per the MOP, rather than the historical (INITARG INITFUNCTION
1061 INITFORM). (reported by Bruno Haible)
1062 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1063 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1064 for more information.
1065 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1066 pathname is a directory pathname.
1067 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1068 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1070 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1071 :METACLASS argument in addition to a class name. (reported by
1072 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1073 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1074 (reported by Bruno Haible; patch from Gerd Moellmann)
1075 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1077 * bug fix: more cleanups to the floating point exception handling on
1078 x86-64 (thanks to James Knight)
1079 * bug fix: the compiler does not try to constant fold calls of
1080 COERCE and BIT-* functions when they return freshly constructed
1081 objects. (reported by Nikodemus Siivola and Paul Dietz)
1082 * optimization: improved the MIPS versions of generic (in the
1083 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1084 * optimization: direct conversion of (unsigned-byte 32) to floats on
1085 the PowerPC platform.
1086 * optimization: structure instances with raw slots now use less
1087 memory, and probably show better memory locality. (thanks to
1089 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1090 are treated as requests for stack allocation on the x86-64,
1091 Alpha-32, PPC and SPARC platforms.
1092 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1093 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1094 (thanks to Tony Martinez)
1096 ** gcing a dead thread can no longer lead to lockups
1097 ** threads block signals until they are set up properly
1098 ** errno is no longer shared by threads
1099 ** interrupt-thread restores the eflags register on x86
1100 ** fixed some lockups due to gc/thread interaction
1101 * fixed some bugs revealed by Paul Dietz' test suite:
1102 ** invalid dotted lists no longer raise a read error when
1103 *READ-SUPPRESS* is T
1104 ** signal an error if a symbol that names a declaration is used
1105 as the name of a type, or vice versa
1106 ** allow using the (declare (typespec var*)) abbreviation for
1107 (declare (type typespec var*)) with all type specifiers
1108 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1109 ** always return NIL from PPRINT-POP when OBJECT is NIL
1110 ** don't signal errors when pretty-printing malformed LABELS,
1111 FLET or MACROLET forms
1112 ** declarations in a DOLIST body are also in scope for the
1114 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1116 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1119 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1120 * fixed cross-compiler leakages that prevented building a 32-bit
1121 target with a 64-bit host compiler.
1122 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1123 opened with :IF-EXISTS :OVERWRITE.
1124 * fixed bug 281: error for an invalid qualifier in a short-form method
1125 combination method is not signalled until the faulty method is called.
1126 * bug fix: iteration variable type inferrer failed to deal with open
1127 intervals. (reported by Alan Shields)
1128 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1129 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1130 or more runtime options were provided to the sbcl binary.
1131 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1132 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1133 * optimization: there's now a fast-path for fixnum arguments in the
1134 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1135 * optimization: the code generated on x86-64 is more compact thanks
1136 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1137 * bug fix: floating-point exception handling now partly works on
1138 x86-64. (Thanks to James Knight)
1139 * improvement to the MIPS backend from Thiemo Seufer: C-style
1140 64-bit long long arguments and return values to alien functions
1142 * fixed some bugs revealed by Paul Dietz' test suite:
1143 ** the type-error signalled from WARN has a filled-in DATUM slot.
1144 ** the type-error required when a stream is not associated with
1145 a file has the stream as its datum.
1146 ** type-errors on single-floats on x86-64 no longer have
1147 :INVALID-OBJECT as the datum
1148 ** the type-errors signalled for invalid function names now have
1149 a correct expected type
1150 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1151 for structures defined with a :TYPE. Documentation strings for
1152 typed structures are no longer immediately discarded
1153 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1154 broadcast streams and synonym streams. FILE-LENGTH now also works
1155 on broadcast streams.
1157 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1158 * incompatible change: the --noprogrammer option, deprecated since
1159 version 0.7.5, has been removed. Please use the equivalent
1160 --disable-debugger option instead.
1161 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1163 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1164 inoperational for a while now) has been completely removed.
1165 * null lexical environments are now printed as #<NULL-LEXENV>,
1166 significantly reducing the amount of clutter in typical backtraces.
1167 * documentation on weak pointers, finalization, and after GC hooks
1168 has been added to the manual.
1169 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1170 as other specialized array types, is much faster. SUBSEQ and
1171 COPY-SEQ on such arrays have also been sped up.
1172 * optimization: EQL is now more efficient when at least other argument
1173 is known to be of type (OR FIXNUM (NOT NUMBER)).
1174 * fixed bug: the runtime is now less vulnerable to changes in the
1175 size of the SBCL object on OS X, and virtual memory is reserved for
1176 all spaces that need to be at a fixed address.
1177 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1178 * fixed bug: finalizers and after GC hooks that cause consing are now
1180 * fixed bug: compiler error messages and summaries are now printed to
1181 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1182 * fixed inference of the upper bound of an iteration variable.
1183 (reported by Rajat Datta).
1184 * fixed bug 211e: calling local functions with duplicated constant
1185 keyword argument no longer causes a bogus style warning about an
1187 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1188 ftype declaration to be disregarded. (reported by Dave Roberts)
1189 * fixed bug 373: caused by erronous compilation of references to alien
1190 variables in the runtime on ppc/darwin.
1191 * fixed bug 376: CONJUGATE type deriver.
1192 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1193 (reported by Baughn on #lisp)
1194 * a cross-compiler bug on non-x86 platforms has been identified and
1195 fixed. (thanks to Bruno Haible)
1196 * improvements to the MIPS runtime code for increased stability.
1197 (thanks to Thiemo Seufer)
1198 * increased the maximimum compact environment size to allow
1199 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1200 * partial workaround for bug 135: don't name the function we're
1201 calling for hairy cases of EVAL, so as not to accumulate one environment
1202 entry per call to EVAL. (reported by Kevin Reid)
1203 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1204 * optimization: type testing for non-vector arrays should be faster.
1205 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1207 * bug fix: setting 31st element of a bit vector to zero did not work
1209 * bug fix: redefining a class definition which failed due to a
1210 previous accessor / function clash now works (but see BUGS entry
1211 #380 for more problems in this area). (thanks to Zach Beane)
1212 * the long form of DEFINE-METHOD-COMBINATION disables method group
1213 checking when given a single method group with pattern *. (thanks
1215 * on x86 compiler supports stack allocation of results of simple
1216 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1217 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1218 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1219 * fixed some bugs related to Unicode integration:
1220 ** the restarts for recovering from input and output encoding
1221 errors only appear when there is in fact such an error to
1223 * fixed some bugs revealed by Paul Dietz' test suite:
1224 ** MISC.549 and similar: late transformation of unsafe type
1225 assertions into derived types caused unexpected code
1227 ** SCALE-FLOAT type deriver is less wrong.
1228 ** type derivers for EXP, LOG and similar functions compute result
1229 types for complex arguments better.
1230 ** (MISC.563) CONJUGATE type deriver works for very restricted
1232 ** out-of-line type testers for character strings are available.
1233 ** EQUAL compiler transform understands specialness of objects
1235 ** accessing double-floats stored on the stack now works on x86-64.
1236 ** debugger internals could sometimes create invalid lispobjs,
1237 resulting in GC crashes.
1238 ** MISC.548: type check weakening can convert required type into
1240 ** initialization forms for bindings are not in scope of free special
1243 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1244 * incompatible change: thread support for non-NPTL systems has
1245 been removed - locking is buggy and unreliable. A threaded
1246 SBCL build will now warn at startup and refuse to create
1247 new threads, unless futex support is detected in the kernel
1248 * incompatible change: the top level REPL now has only an ABORT
1249 restart associated with it, not TOPLEVEL and ABORT as it used to.
1250 TOP and TOPLEVEL are now available as debugger commands for
1251 returning to the top level.
1252 * incompatible change: forms evaluated in the REPL now use the
1253 global optimization policy.
1254 * incompatible change: user- and system-initialization files are
1255 no longer processed with LOAD, but by READ and EVAL; hence the
1256 global optimization policy, startup package, readtable, etc,
1258 * The .fasl file format number has been incremented because of
1259 various incompatible changes.
1260 * internal entry point details and argument counts no longer appear
1261 in backtraces unless explicitly requested by setting
1262 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1263 * built-in and standard functions no longer have names like "top
1264 level local call to FOO".
1265 * fixed bug 32: functions defined in non-null lexical environments
1266 now have more legible printed representation
1267 * fixed bug 33: functions defined in non-null lexical environemnts
1268 are now more amenable to inspection by INSPECT.
1269 * workaround for bug 354: XEPs no longer appear in backtraces unless
1270 explicitly requested.
1271 * fixed bug: receiving the signal which results from attempting to
1272 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1273 SIGBUS on other *BSDs) on architectures where the C stack is also
1274 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1275 to debugger confusion or wild instability. (reported by Cheuksan
1277 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1278 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1279 (reported by Lutz Euler)
1280 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1281 compiled code. (reported by Paul Dietz)
1282 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1283 specialized parameter to a method to something that is not TYPEP
1284 the specializer is now possible.
1285 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1286 face of package deletion.
1287 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1288 pathname. (reported by David Tolpin on c.l.l)
1289 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1290 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1291 STRING-STREAM and FILE-STREAM, respectively.
1292 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1293 than previously; in addition, its test suite is now run on build.
1294 (thanks to Luke Gorrie)
1295 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1296 * optimization: added a immediate representation for single-floats
1298 * fixed some bugs revealed by Paul Dietz' test suite:
1299 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1300 ** attempting to create a package with a colliding nickname causes
1301 correctable errors to be signalled.
1302 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1303 ** several x86-64 backend bugs related to sign-extension of immediate
1306 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1307 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1308 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1310 * fixed loading of multiply forward-referenced layouts.
1311 (thanks to Cheuksan Wang)
1312 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1313 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1314 (see COMPILE-FILE documentation for details of :PRINT option)
1315 * fixed bugs 19 and 317: fixed-format floating point printing is
1316 more accurate. This also fixes a bug reported by Adam Warner
1317 related to the ~@F format directive.
1318 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1320 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1321 dispatch table if the from-char is a dispatch macro character.
1322 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1323 by the wordsize no longer ignore the last word. (reported by Lutz
1325 * fixed bug in type checking of dynamic-extent variables. (reported
1327 * optimization: sequence traversal functions use their freedom to
1328 coerce function designators to functions.
1329 * optimization: code with many calls to CLOS methods specialized on
1330 CLOS classes has had redundant type checks removed.
1331 * optimization: added declarations to speed up operations that access
1332 the internal character database (for example STRING-UPCASE)
1333 * optimization: comparison operations between floats and sufficiently small
1334 fixnums no longer create extra rationals
1335 * fixed some bugs related to Unicode integration:
1336 ** portions of multibyte characters at the end of buffers for
1337 character-based file input are correctly transferred to the
1338 start of the buffer at the next read.
1339 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1340 passing it through to OPEN.
1341 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1342 argument given to internal calls to OPEN.
1343 ** the built-in comment readers (introduced by character sequences
1344 ";" and "#|") are more forgiving to encoding errors; they will
1345 STYLE-WARN and then attempt to resync the stream at a character
1346 boundary. (thanks to Teemu Kalvas)
1347 * fixed some bugs in the x86-64 port:
1348 ** Negative short int return values from c-calls are sign-extended
1350 ** The stack is aligned to 16-bytes for c-calls, as required by
1351 the ABI. (thanks to Cheuksan Wang)
1352 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1353 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1354 for enums. (thanks to Vincent Arkesteijn)
1355 ** Multiple small optimizations and bugfixes for floating point
1357 * fixed some bugs revealed by Paul Dietz' test suite:
1358 ** Space, Tab, Linefeed, Return and Page have the invalid
1359 secondary constituent character trait.
1360 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1362 ** WITH-INPUT-FROM-STRING should only update the index place on
1364 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1365 works more reliably.
1366 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1367 with one-dimensional arrays with fill pointers.
1368 ** TYPE-OF failed on a complex with an integer realpart and a
1370 ** compiler failure during type inference for the code of form
1371 (IF (EQL X (THE ...)) ...) (MISC.535).
1373 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1374 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1375 platform. The port supports 61-bit fixnums, large memory spaces
1376 and reloading shared object files.
1377 * enhancement: saving cores with foreign code loaded is now
1378 supported on ppc/Darwin in addition to the previously supported
1380 * enhancement: the statistical profiler now walks deeper into the
1381 call stack for more meaningful call-graphs and accrued time
1382 reports (x86/x86-64 only). It also now reports time spent in
1384 * enhancement: it is now possible to trace most individual methods
1385 of a generic function in addition to tracing the generic function
1387 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1389 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1390 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1391 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1392 produces an error. (thanks to Vincent Arkesteijn)
1393 * bug fix: NAMESTRING on pathnames with :WILD components in their
1394 directories works correctly. (thanks to Artem V. Andreev)
1395 * fixed bug 125: compiler preserves identity of closures. (reported
1397 * bug fixed: functions with &REST arguments sometimes failed with
1398 "Undefined function" when compiled with (DEBUG 3). (reported by
1400 * bug fix: overflow during compiling of setting element of a bit
1401 vector with constant index and value. (reported by Timmy Douglas)
1402 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1403 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1404 * on x86 compiler partially supports stack allocation of dynamic-extent
1406 * GO and RETURN-FROM do not check the extent of their exit points
1407 when compiled with SAFETY 0.
1408 * fixed some bugs related to Unicode integration:
1409 ** encoding and decoding errors are now much more robustly
1410 handled; it should now be possible to recover even from invalid
1411 input or output to the terminal. (thanks to Teemu Kalvas)
1412 ** provided a first cut at implementing STRING-TO-OCTETS and
1413 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1414 ** altered the SB-MD5 contributed module slightly, changing the
1415 interface just enough for it to be supportable for builds where
1416 lisp characters are not eight bits.
1417 * fixed some bugs revealed by Paul Dietz' test suite:
1418 ** the FORMATTER-generated functions for ~V[ conditionals require
1419 the correct number of arguments.
1420 ** READ-FROM-STRING returns the mandated second value when applied
1421 to displaced strings.
1422 ** the #\Rubout and #\Backspace characters are treated as invalid
1423 constituent characters by the tokenizer.
1425 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1426 * new feature: reloading changed shared object files with
1427 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1428 * new feature: references to foreign variables and functions
1429 can now be compiled and loaded before the corresponding shared
1430 object file is loaded, as long as the foreign definitions are
1431 available at runtime.
1432 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1433 supported. (thanks to Dan Debertin)
1434 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1435 just keywords. (thanks to Vincent Arkesteijn)
1436 * fixed bug #331: structure-class instances corresponding to
1437 DEFSTRUCT forms are now created eagerly.
1438 * fixed bug #345: backtraces from calls to undefined functions work
1439 on x86 as well. Related bug #61 is now also partially fixed on x86
1440 (backtraces from throws to unknown catch tags.)
1441 * bug fix: lambda-list parsing is now stricter vrt. order and number
1442 of lambda-list keywords.
1443 * bug fix: as specified by AMOP, an error is signalled if a
1444 class-option appears multiple times in a DEFCLASS form. (reported
1446 * bug fix: RANDOM can be compiled when the compiler derives the type
1447 of its numeric argument as a disjoint set of small integers.
1448 (reported by Paul Dietz)
1449 * bug fix: starting a core saved with shared objects loaded when
1450 those objects are not available no longer causes threaded SBCL to
1451 hang. (reported by Sean Ross)
1452 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1454 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1455 argument, which must always be NIL. (reported by Kalle Niemitalo)
1456 * bug fix: printing 1.0d+23 no longer results in an error.
1457 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1458 * bug fix: structure slot setters preserve evaluation order. (thanks
1460 * fixed some bugs related to Unicode integration:
1461 ** RUN-PROGRAM can allow its child to take input from a Lisp
1462 stream. (reported by Stefan Scholl)
1463 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1464 locales. (reported by Ken Causey)
1465 * fixed some bugs revealed by Paul Dietz' test suite:
1466 ** INCF, DECF and REMF evaluate their place form as specified in
1468 ** FORMATTER expands ~{ ~} iteration directives with V or #
1469 parameters correctly.
1470 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1471 consequent uses no arguments correctly.
1472 ** the system has a partial understanding of the (COMPLEX RATIO)
1474 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1477 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1478 * new feature: a build-time option (controlled by the :SB-UNICODE
1479 keyword feature, enabled by default) for building the system with
1480 support for the entire 21-bit character space defined by the
1482 * new feature: the system now has rudimentary external-format
1483 support; the primary user-visible change at this time is that
1484 characters with the high bit set (such as lower-case-e-acute) will
1485 print correctly to a terminal in a UTF-8 environment.
1486 * minor incompatible change: BASE-CHAR no longer names a class;
1487 however, CHARACTER continues to do so, as required by ANSI.
1488 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1489 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1490 has been moved to the SB-EXT package (temporarily re-exported from
1492 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1493 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1495 * fixed bug #351: better error-handlind and reporting for malformed
1497 * fixed bug #350: bignum-printing is now more memory-efficient,
1498 allowing printing of very large bignums, eg. (expt 2 10000000).
1499 (reported by Bruno Haible)
1500 * fixed bug #302: better primitive-type selection for intersection
1502 * fixed bug #308: non-graphic characters now all have names, as
1503 required. (reported by Bruno Haible)
1504 * bug fix: redefining a class with different superclasses now correctly
1505 removes it from the direct-subclasses of its previous superclasses.
1506 (reported by David Morse)
1507 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1508 new value now works. (reported by Bruno Haible)
1509 * bug fix: correct canonicalization of multiple non-standard slot
1510 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1511 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1512 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1513 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1514 now exists, an signals an error.
1515 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1516 during macroexpansion for non-iteration variables as well. (reported
1517 by Bruno Haible for CMUCL)
1518 * bug fix: Cyclic structures and unprintable objects in compiler
1519 messages no longer cause errors. (reported by Bruno Haible)
1520 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1521 and READ-FROM-STRING all now return a primary value of NIL if
1522 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1523 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1524 (reported by Bruno Haible for CMUCL)
1525 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1526 do not designate a condition. (reported by Bruno Haible for
1528 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1529 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1530 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1531 permitting resolution in favour of any of the conflicting symbols.
1532 (reported by Bruno Haible for CMUCL)
1533 * bug fix: EQUAL compiler optimizations is less aggressive on
1534 strings which can potentially compare true despite having distinct
1535 specialized array element types.
1536 * bug fix: unit enumerations can be defined without dividing by
1537 zero. (thanks to Vincent Arkesteijn)
1538 * FORMAT compile-time argument count checking has been enhanced.
1539 (report from Bruno Haible for CMUCL)
1540 * a partial workaround for the bug 262: the compiler does not try to
1541 inline-expand a local function doing RETURN-FROM from a deleted
1542 BLOCK. (thanks to Peter Denno for the bug report and to David
1543 Wragg for the simple test case)
1544 * fixed some bugs revealed by Paul Dietz' test suite:
1545 ** RENAME-PACKAGE allows all package designators as new package
1547 ** constraint propagation and lambda variable substitution are
1548 more cautious in dealing with partially deleted code.
1549 ** compiler sometimes lost reoptimization passes.
1550 ** CERROR, given a condition as condition designator, passes its
1551 remaining arguments to the continue format control without
1553 ** Case-altering FORMAT directives work correctly on non-ASCII
1555 ** The REAL type specifier handles bounds outside the floating
1556 point ranges without signalling FLOATING-POINT-OVERFLOW.
1557 ** Functions with IR1-transformations can create intercomponent
1558 references to global functions.
1559 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1561 ** FORMAT ~R treats a nil value for its first parameter correctly.
1563 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1564 * enhancement: saving cores with foreign code loaded is now
1565 supported on x86/NetBSD and sparc/Linux in addition to the previously
1566 supported platforms.
1567 * bug fix: on some platforms repeated installations caused multiple
1568 copies of HTML documentation to be installed -- should not happen
1569 any more. (reported by Stefan Scholl)
1570 * bug fix: parsing self-recursive alien record types multiple times
1571 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1572 original patch by Helmut Eller for CMUCL)
1573 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1574 (thanks to Richard Kreuter)
1575 * bug fix: defining classes whose accessors are methods on existing
1576 generic functions in other (locked) packages no longer signals
1577 bogus package lock violations. (reported by François-René Rideau)
1578 * bug fix: special variables as DEFMETHOD parameters no longer have
1579 associated bogus type declarations. (reported by David Wragg and
1581 * bug fix: read-write consistency on streams of element-type
1582 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1583 * bug fix: redefiniton of the only method of a generic function with
1584 no DEFGENERIC no longer emits a full WARNING. In addition,
1585 redefinition of generic functions with no DEFGENERIC to an
1586 incompatible lambda list now signals an error. (thanks to Zach
1588 * bug fix: DEFGENERIC now works even when there's a function of the
1589 same name in an enclosing lexical environment. (thanks to Zach
1591 * fixed compiler failure, caused by instrumenting code during
1592 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1593 * optimization: added loop analysis and improved register allocation
1594 * fixed some bugs revealed by Paul Dietz' test suite:
1595 ** POSITION on displaced vectors with non-zero displacement
1596 returns the right answer.
1597 ** (SIMPLE-STRING) is a valid type specifier for sequence
1599 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1601 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1602 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1604 ** The justification version of the FORMAT ~< directive treats
1605 non-zero minpad parameter correctly.
1607 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1608 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1609 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1610 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1611 the supported interface.
1612 * new feature: Single-stepping of code compiled with DEBUG 2 or
1613 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1614 * new feature: saving cores with foreign code loaded is now
1615 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1616 Timothy Moore's work for CMUCL)
1617 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1618 parameters to * instead of NIL if no initform is supplied.
1619 (reported by Johan Bockgaard)
1620 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1621 correctly when FUNCALL appears as the car of the form. Note:
1622 despite this FUNCALL forms are not currently subject to
1623 compiler-macro expansion. (port of Raymond Toy's fix for the
1624 same from CMUCL, reported by Johan Bockgaard)
1625 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1626 (thanks for Teemu Kalvas)
1627 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1628 case instead of terminating the iteration (thanks for Julian
1629 Squires, Sean Champ and Raymond Toy)
1630 * bug fix: incorrect expansion of defgeneric that caused a style
1631 warning. (thanks for Zach Beane)
1632 * bug fix: slot accessor effective method computation works properly
1633 for classes with multiple non-standard applicable methods on
1634 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1635 * on x86 compiler supports stack allocation of results of LIST and
1636 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1637 CMUCL implementation by Gerd Moellmann)
1638 * fixed some bugs revealed by Paul Dietz' test suite:
1639 ** FORMAT strings with both the ~<~:;~> form of the justification
1640 directive and pretty-printing directives cause an error.
1642 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1643 * incompatible change: the internal functions
1644 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1645 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1646 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1647 instead of the old functions.
1648 * new feature: on platforms where "dladdr" is available foreign
1649 function names now appear in backtraces. (based on Helmut Eller's
1651 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1652 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1654 * documentation: ASDF manual is now distributed with SBCL.
1655 * bug fix: SBCL can now load its contributed modules with REQUIRE
1656 even if the system-provided entries have been removed from
1657 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1659 * bug fix: backtraces involving undefined functions or assembly
1660 routines are more informative. (thanks to Brian Downing)
1661 * bug fix: mutually referent alien structures now work correctly.
1662 (reported by Rick Taube)
1663 * bug fix: structures defined by WITH-ALIEN can be referred to
1664 within other definitions in the same WITH-ALIEN.
1665 * bug fix: division operators (MOD, TRUNCATE and the like) with
1666 constant zero divisors and integer dividends no longer generate
1668 * bug fix: provide default methods for INPUT-STREAM-P and
1669 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1670 * bug fix: improve the detection and resolution of MOP metacycles
1671 (where effective method computation depends on the generic function
1672 having its effective method computed). (reported by Bruno Haible)
1673 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1674 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1675 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1676 * bug fix: pathnames with relative directory components can be
1677 represented relative to default pathnames.
1678 * optimization: in taking the GCD of bignums, reduce the two bignums
1679 to approximately the same size (using Euclid's algorithm) before
1680 applying the more sophisticated binary GCD. (thanks to Juho
1682 * optimization: COUNT on bitvectors now operates word-at-a-time.
1683 * optimization: ASH with a positive, but not necessarily constant,
1684 (leftwards) shift, when in a modular context, is compiled to a
1686 * fixed some bugs revealed by Paul Dietz' test suite:
1687 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1688 if the corresponding argument is NIL.
1689 ** FORMAT directives accepting a minpad parameter treat negative
1691 ** literal commas as character parameters to FORMAT directives are
1693 ** literal spaces directly after ~<Newline> directives within a
1694 format-logical-block (~:< ~@:>) do not induce :FILL-style
1695 conditional newlines.
1696 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1697 as specified: it no longer includes conditional newlines).
1698 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1700 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1701 :PER-LINE-PREFIX argument does not evaluate to a string.
1703 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1704 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1705 the manual for details; for now, package locks can be disabled by
1706 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1707 if no major problems are found then it is likely that they will be
1708 compiled in unconditionally.
1709 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1710 unsupported operators on all platforms. To load a shared library
1711 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1712 object file, link it into a shared library outside of SBCL and
1713 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1715 * bug workaround: ROOM T and the GC were not working together
1716 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1717 reporting facility aren't preserved. That reporting has been
1718 disabled, so now until and unless someone figures out how to make it
1719 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1720 * minor incompatible change: as threatened around sbcl-0.8.0, the
1721 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1722 The symbols are also exported from SB-PCL for backwards
1723 compatibility, but more so than before SB-PCL should be treated as
1724 an implementation-internal package.
1725 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1727 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1728 It is known as of this release not to work on the Alpha, however.
1729 * fixed bug #167: errors signalled due to illegal syntax in method
1730 bodies are now more legible.
1731 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1732 designators and can hence be used with TYPEP.
1733 * fixed bug #333: CHECK-TYPE now ensures that the type error
1734 signalled, if any, has the right object to be accessed by
1735 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1736 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1737 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1739 * fixed bug #334: programmatic addition of slots using specialized
1740 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1741 and :ALLOCATION :CLASS slots.
1742 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1743 float exponents. (rereported by Peter Seibel)
1744 * fixed a bug: #\Space (and other whitespace characters) are no
1745 longer considered to be macro characters in standard syntax by
1746 GET-MACRO-CHARACTER.
1747 * fixed bug: initialization of condition class metaobjects no longer
1748 causes an instance of the condition to be created. (reported by
1750 * fixed bug: it is now possible to have slots such that
1751 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1752 description is neither :INSTANCE nor :CLASS.
1753 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1754 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1755 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1756 non-standard SLOT-DEFINITION-ALLOCATION values.
1757 * fixed bugs: various race conditions handling exiting threads.
1758 CL-PPCRE's thread torture test now passes.
1759 * fixed arguably-a-bug: GC time proportional to number of threads in
1760 system even when most of them are idle
1761 * optimization: improved performance of BIT and SBIT on bit-vectors.
1762 * .fasl file incompatibility: The fasl file version number has
1763 been incremented because of changes associated with package locks.
1765 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1766 * minor incompatible change: the system no longer provides
1767 optimization hints (conditions which are TYPEP
1768 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1769 mode; these hints are still emitted when higher SPEED optimization
1771 * new contrib module: a sampling profiler (profiling by statistical
1772 sampling, rather than by instrumenting functions) is available as
1773 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1774 * the behaviour of the standard function ED is now customizeable by
1775 third parties through a hook variable: see ED's documentation
1776 string for information on the protocol.
1777 * the compiler no longer emits efficiency notes for (FUNCALL X)
1778 when the type of X is uncertain under default optimization
1780 * fixed bug 276: mutating a binding of a specialized parameter to a
1781 method to something that is not TYPEP the specializer is now
1783 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1784 exhibit the required behaviour on the x86 platform. (thanks to
1785 Peter van Eynde, Eric Marsden and Bruno Haible)
1786 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1787 even for difficult arguments. (reported by Peter Graves)
1788 * fixed bug 141a: the backquote printer now descends quoted
1790 * fixed another bug in backquote printing: no more destructive
1791 modification of the form's list structure. (reported by Brian
1793 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1794 move between its address being taken and the call to
1795 interrupt_thread, fixing a crashing race condition.
1796 * the SB-POSIX contrib implementation has been adjusted so that it
1797 no longer exhibits ridiculously poor performance when constructing
1798 instances corresponding to C structs.
1800 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1801 * minor incompatible change: the sb-grovel contrib now treats C
1802 structures as alien (in the sense of SB-ALIEN) objects rather than
1803 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1804 has implications for memory management of client code
1805 (sb-grovel-returned objects must now be manually managed) and for
1806 type safety (alien objects now have full types).
1807 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1808 used to control emission of compiler diagnostics, rather than the
1809 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1810 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1811 quality should be considered deprecated.
1812 * install.sh now installs the user manual as well
1813 * (not quite a new documentable feature, but worth considering in
1814 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1815 beginnings of a semantically meaningful condition hierarchy is
1816 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1818 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1819 designator as the defaults argument.
1820 * fixed bug: Displaced arrays whose displaced-to array has become
1821 too small now cause ARRAY-DIMENSION to signal an error, providing
1822 for safer bounds-checking. (reported by Bruno Haible)
1823 * fixed bug: DEFCLASS slot definitions with identical :READER and
1824 :WRITER names now signal a reasonable error. (reported by Thomas
1826 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1828 * fixed bug: Invalid slot specification errors now print correctly.
1829 (thanks to Zach Beane)
1830 * fixed bug 320: Shared to local slot value transfers in class
1831 redefinitions now happen corrently from superclasses as
1832 well. (reported by Bruno Haible)
1833 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1835 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1836 improper lists correctly. (reported by Bruno Haible)
1837 * fixed bug 313: source-transform for <fun-name> was erroneously
1838 applied to a call of a value of a variable with name <fun-name>.
1839 (reported by Antonio Menezes Leitao)
1840 * fixed bug 307: The obsolete instance protocol ensures that
1841 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1842 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1843 remove dead unknown-values globs from the middle of the stack.
1844 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1845 easier to resume long computations after using *BREAK-ON-SIGNALS*
1846 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1847 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1848 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1849 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1850 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1851 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1853 * fixed a (fairly theoretical) bug in string printing: if
1854 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1855 does not have array-element-type equal to the most general string
1857 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1858 function name. (thanks to Nikodemus Siivola)
1859 * fixed bug: compile-time format string checker failed on
1860 non-closed ~{. (reported by Thomas F Burdick)
1861 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1862 don't warn on legal loop code involving more than one aggregate
1863 boolean. (thanks to Nikodemus Siivola)
1864 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1865 in keyword position to MAKE-PACKAGE was not being reported as
1867 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1868 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1870 * fixed bugs 280 and 312: the checking for multiple definitions in a
1871 file is less likely to become confused by uses of inline
1873 * fixed bug: the #S reader macro performs the keyword coercion
1874 specified for slot names. (reported by Kalle Niemitalo)
1875 * fixed bug: lambda lists may contain symbols whose names start with
1876 & but are not lambda-list-keywords; their occurrence triggers a
1878 * fixed bug 321: define-method-combination argument lambda lists do
1879 not cause invalid code to be generated when &optional or &aux
1880 variables have default values. (reported by Bruno Haible)
1881 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1882 have CLOS classes; this appears also to have fixed reported
1883 instability in stack exhaustion detection.
1884 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1885 MUFFLE-WARNING when no associated restart is present is now
1887 * optimization: rearranged the expansion of various defining macros
1888 so that each expands into only one top-level form in a
1889 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1891 * optimization: used a previously little-used slot in symbols to
1892 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1894 * fixed some bugs revealed by Paul Dietz' test suite:
1895 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1897 ** Fixed an optimization bug related to inheritance of initforms
1898 from local to shared slots.
1899 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1900 ** CAST optimizer forgot to flush argument derived type.
1901 ** print/read consistency on floats is now orders of magnitude
1902 more likely. (thanks also to Bruno Haible for a similar report
1904 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1905 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1906 for objects of type REAL. Make it so.
1907 ** SXHASH is specified (infelicitously) to respect similarity,
1908 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1909 it so. (thanks to Markus Ziegler)
1910 ** on the Alpha, the compiler succeeds in compiling functions
1911 returning a known number of arguments greater than 63.
1912 ** fixed handling of invalid NIL arguments in keyword position
1914 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1916 ** provided a readably-printable representation for RANDOM-STATE
1918 ** ensured that pathnames, where they have a namestring, always
1919 print using #P"..." syntax.
1921 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1922 * Support for the forthcoming 2.0 version of the NetBSD kernel
1923 running on x86 hardware has been added. (thanks to Perry
1924 E. Metzger most immediately, and others for their past work)
1925 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1926 current release version; SBCL's OpenBSD support had been broken
1927 since about the time of OpenBSD's switch to ELF binary format.)
1928 * [placeholder for DX summary]
1929 ** user code with &REST lists declared dynamic-extent, under high
1930 speed or space and low safety and debug optimization policy.
1931 * The manual has been converted to Texinfo format and the debugger
1932 chapter from the cmucl manual has been added.
1933 * A facility has been added to extract documentation strings from
1934 sbcl and store them as Texinfo-formatted snippets for inclusion in
1935 the manual (via Texinfo's @include directive)
1936 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1937 the test case to Dave Roberts)
1938 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1939 pointers. (reported by Sean Ross)
1940 * bug fix: PROFILE output is printed nicely even for large numerical
1941 values. (thanks to Zach Beane)
1942 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1943 greater than 32 handle EOF correctly.
1944 * bug fix: on X86 an immediate argument of the IMUL instruction is
1945 correctly printed in disassembly. (thanks to Lutz Euler)
1946 * bug fix: class slots in redefined classes preserve their old
1947 values. (thanks to Bruno Haible and Nikodemus Siivola)
1948 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1949 succeeds. (reported by Marco Baringer)
1950 * fixed some bugs revealed by Paul Dietz' test suite:
1951 ** READ-SEQUENCE now works on ECHO-STREAMs.
1952 ** RATIONALIZE works more according to its specification. (thanks
1955 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1956 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1957 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1958 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1959 should matter to you only if you rebind the printer control
1960 variables and then find you want different bindings in the
1961 debugger than in the ordinary execution of your program.)
1962 * The runtime build system has been tweaked to support building
1963 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1964 assembler and linker. (thanks to Nikodemus Siivola)
1965 * Unbound, undefined, undeclared variables now trigger full
1966 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1967 more often programmer error than deliberate exploitation of undefined
1969 * optimization: the hash algorithm for strings has changed to one
1970 that is less vulnerable to spurious collisions. (thanks to Juho
1972 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1973 needless bounds checking. (thanks to Juho Snellman)
1974 * optimization: implemented multiplication as a modular
1975 (UNSIGNED-BYTE 32) operation on the PPC backend.
1976 * fixed some bugs revealed by Paul Dietz' test suite:
1977 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1979 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1980 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1981 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1983 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1985 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1987 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1988 their output stream on EOF from read.
1989 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1990 have been read to end-of-file.
1991 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1993 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1994 description of determination of which consecutive characters
1996 ** Printing the "Space" character with escaping on now yields "#\\ ",
1997 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1998 ** Reading floating-point numbers with *READ-BASE* set to a number
1999 less than 10 works correctly.
2000 ** Reading floating-point numbers with *READ-BASE* set to a number
2001 more than 10 works correctly.
2002 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2003 the readtable currently in effect.
2005 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2006 * minor incompatible change: parsing of namestrings on a physical
2007 (Unix) host has changed; numbers after the final #\. in a
2008 namestring are no longer interpreted as a version field. This is
2009 intented to be largely invisible to the user, except that the
2010 meaning of the namestring "*.*.*" has changed: it now refers to a
2011 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2012 should usually be replaced by
2013 (make-pathname :name :wild :type :wild :version :wild)
2014 with the added benefit that this is more likely to be portable.
2015 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2016 option to OPEN now signals an error if the file being opened
2017 exists; this may have an impact on existing code.
2018 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2019 signals received, on the PowerPC platforms (both Linux and
2020 Darwin). (thanks to Pierre Mai for pointing out the location of
2022 * several fixes on OS X: The system now builds and runs cleanly on
2023 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2024 (thanks to Brian Mastenbrook)
2025 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2026 second-resolution: integer multiples of 1/3600 between -24 and 24.
2027 (thanks to Vincent Arkesteijn)
2028 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2029 type when called with 1 argument; PEEK-CHAR checked type of
2030 PEEK-TYPE only after having read first character from a
2031 stream. (reported by Peter Graves)
2032 * bug fix: the garbage collector now has much better locality
2033 behaviour, and in particular no longer treats all memory as being
2034 exhausted when many small objects point to each other in a deeply
2036 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2037 recognized as being TYPEP their class.
2038 * bug fix: the PUSHNEW documentation string has been corrected.
2039 (thanks to Vincent Arkesteijn)
2040 * bug fix: defaulting of the value for the last of an atypically
2041 large number of multiple values being bound was not being
2042 performed correctly on the Alpha or PPC platforms
2043 * optimization: implemented multiplication as a modular
2044 (UNSIGNED-BYTE 32) operation on the x86 backend.
2045 * optimization: SEARCH on simple-base-strings can now be open-coded.
2046 (see also contrib/compiler-extras.lisp for inspiration for
2047 teaching the compiler about the Boyer-Moore algorithm).
2048 * value, returned by MAX (and MIN) called with several EQUALP, but
2049 not EQL, arguments now does not depend on compiler settings.
2050 * fixed some bugs revealed by Paul Dietz' test suite:
2051 ** in stack analysis liveness information is propagated from
2052 non-local entry points.
2053 ** pathwise CAST removing failed when the CAST node did not start
2055 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2056 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2058 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2059 host is already defined.
2060 ** RENAME-FILE works on streams instead of signalling an internal
2062 ** PEEK-CHAR uses the current readtable when determining whether
2063 or not a character is whitespace.
2064 ** MERGE-PATHNAMES handles the case when the pathname does not
2065 specify a name while the default-pathname specifies a version.
2066 ** Pathnames now stand a better chance of respecting print/read
2068 ** Attempting to use standardized file system operators with a
2069 pathname with invalid :DIRECTORY components signals a
2071 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2073 ** DIRECTORY on logical pathnames is more correct.
2074 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2075 signal a TYPE-ERROR if their argument is not a stream.
2076 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2077 designator argument does not designate a stream.
2078 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2079 examining the synonym.
2080 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2082 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2083 element-type larger than ([UN]SIGNED-BYTE 32).
2085 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2086 * When built with the :SB-FUTEX feature, threaded builds now take
2087 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2088 for faster/more reliable mutex and condition variable support.
2089 * Incompatible change (but one you probably shouldn't have been using
2090 anyway): the interface and code for arbitrating between multiple
2091 threads in the same user session has been redesigned.
2092 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2093 error when called without an explicit environment argument.
2094 (thanks to Nikodemus Siivola)
2095 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2096 confused about their position. (thanks to Adam Warner and Gerd
2098 * bug fix: Pretty printing backquoted forms with unquotations in the
2099 argument list position of various code constructs such as LAMBDA
2100 now works correctly. (reported by Paul Dietz)
2101 * bug fix: Pretty printing unquotations no longer loses all
2102 stream position information.
2103 * optimization: performance of string output streams is now less
2104 poor for multiple small sequence writes.
2105 * optimization: performance of CSUBTYPEP in the presence of complex
2106 expressions involving CONS and NOT many times has been improved.
2107 (reported by Paul Dietz)
2108 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2110 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2111 streams for the REPL, for the debugger, and so on. (thanks to
2113 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2114 * fixed some bugs revealed by Paul Dietz' test suite:
2115 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2116 is ignored if it is too small, rather than propagating through
2117 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2118 ** extremely complex negations of CONS types were not being
2119 sufficiently canonicalized, leading to inconsistencies in
2121 ** VALUES tranformer lost derived type.
2123 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2124 * fixed a bootstrapping bug: the build process no longer assumes
2125 that the various BOOLE-related constants have the same value in
2126 host and target lisps. (noted by Paul Dietz' test suite on an
2127 SBCL binary built from CLISP)
2128 * The system can now be dynamically linked on the MIPS platform,
2129 which enables dynamic loading of foreign code from Lisp. (thanks
2130 to Ralf Baechle for discussions on the MIPS ABI)
2131 * The system now records debugging information for its own source
2132 files in a filesystem-position-independent manner, relative to
2133 the "SYS" logical host.
2134 * fixed a compiler bug: MV-LET convertion did not check references
2135 to the "max args" entry point. (reported by Brian Downing)
2136 * tweaked disassembly notes to be less confident about proclaiming
2137 some instruction as an LRA. (thanks to Brian Downing)
2138 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2139 now each have their own history, command character, and other
2140 characteristics. (thanks to David Lichteblau)
2141 * fixed some bugs revealed by Paul Dietz' test suite:
2142 ** compiler failure in compiling LOGAND expressions including a
2144 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2145 shift greater than 32.
2146 ** FLUSH-DEST did not mark blocks for type check regeneration.
2147 ** HANDLER-CASE failed to accept declarations in handler clauses
2148 in some circumstances.
2150 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2151 * New code in contrib/sb-introspect (still probably not entirely
2152 stable yet) provides some support for smart Lisp development
2153 environments like SLIME.
2154 * The conditions signalled for errors occurring when loading .fasl
2155 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2156 in a way which should help ASDF recover gracefully.
2157 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2158 stuff has been cleaned up. If you code contrib/ stuff, this might
2159 affect you, and you can look at contrib/README, contrib/STANDARDS,
2160 and/or the 0.8.4.27 diff to check.
2161 * In full calls the compiler now does not generate checks for declared
2162 argument types for all arguments.
2163 * various threading fixes
2164 ** and some experimental patches which didn't make it into
2165 the main tree for this release, but which are shipped in
2166 contrib/experimental-thread.patch as a possible fix for some
2167 failures (deadlock, spinning...) in GC-intensive multithreaded
2169 * fixed PPC build problem (source code incompatibility of different
2170 library versions): added offsetof() hackery which attempts to divine
2171 where glibc maintainers put uc_mcontext today
2172 * fixed bug 282: compiler does not trust type assertions while passing
2173 arguments to a full call.
2174 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2175 &OPTIONAL VALUES type parameter.
2176 * fix bug 214: algorithm for noting rejected templates is now more
2177 similar to that of template seletion. (also reported by rydis on
2179 * fixed bug 141b: printing backquoted information readably and prettily
2180 inserts a space where necessary.
2181 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2182 (CHARACTER 10) are now reported as errors, rather than propagated
2183 as unknown types. (reported by piso on #lisp)
2184 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2185 with values NIL and :ERROR. (thanks to Milan Zamazal)
2186 * fixed bug 191c: CLOS now does proper keyword argument checking as
2187 described in CLHS 7.6.5 and 7.6.5.1.
2188 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2189 longer raise an error; further, using a list as a for-as-arithmetic
2190 counter now raises a meaningful error.
2191 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2192 now understood by sequence creation functions such as MAKE-SEQUENCE
2194 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2195 a STRING-INPUT-STREAM.
2196 * compiler enhancement: SIGNUM is now better able to derive the type
2198 * type declarations inside WITH-SLOTS are checked. (reported by
2200 * fixed some bugs revealed by Paul Dietz' test suite:
2201 ** incorrect optimization of TRUNCATE for a positive first
2202 argument and negative second.
2203 ** compiler failure in let-convertion during flushing dead code.
2204 ** compiler failure while deriving type of TRUNCATE on an
2205 interval, containing 0.
2206 ** ASH of a negative bignum by a negative bignum count now returns
2208 ** intersection of CONS types now canonicalizes properly, fixing
2209 inconsistencies in SUBTYPEP.
2211 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2212 * incompatible change: The --disable-debugger command line
2213 option now clobbers the debugger at a more fundamental
2214 level, by redefining #'INVOKE-DEBUGGER instead of by
2215 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2216 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2217 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2218 enter the debugger REPL and then suffer recursive errors
2219 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2220 while under the new system, BREAK in --disable-debugger mode
2221 terminates the system just as an unhandled error would.
2222 * fixed compiler performance when processing loops with a step >1;
2223 * bug fix: DOCUMENTATION now retrieves generic function
2224 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2225 support has been systematized, and now supports the methods
2226 specified by ANSI, along with a default method and a method for
2227 slot documentation. (reported by Nathan Froyd)
2228 * bug fix: effective methods associated with a generic function are
2229 no longer cached over a change of that generic function's method
2230 combination. (reported by Andreas Fuchs)
2231 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2232 names a non-existent directory. (thanks to Andreas Fuchs)
2233 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2234 the PPC platform, lead to stack corruption; this has been fixed.
2235 (reported by Rainer Joswig)
2236 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2237 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2238 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2239 arithmetic in general, is now much more reliable on the PPC
2241 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2242 platform now returns the right answer.
2243 * optimization: restored some effective method precomputation in
2244 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2245 precomputation is now tunable.
2246 * optimization: compiler-internal data structure use has been
2247 reviewed, and changes have been made that should improve the
2248 performance of the compiler by about 20%.
2249 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2250 simple-base-strings and simple-bit-vectors is improved.
2251 * optimization: the optimization of 32-bit logical and arithmetic
2252 functions introduced in version 0.8.3 on the x86 has been
2253 implemented on the mips, ppc and sparc platforms; an
2254 implementation of the same facility, but for 64-bit arithmetic,
2255 has been added for the alpha.
2256 * microoptimization: the compiler is better able to make use of the
2257 x86 LEA instruction for multiplication by constants.
2258 * bug fix: in some situations compiler did not report usage of
2259 generic arithmetic in (SPEED 3) policy.
2260 * bug 145b fix: compiler used wrong type specifier while converting
2261 MEMBER-types to numeric.
2262 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2264 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2265 index argument now works properly on the Alpha platform.
2266 * bug fix: floating point exception treatment on the Alpha platform
2268 * bug fix: FILE-POSITION works much better on string input and
2269 output streams. (thanks to Nikodemus Siivola)
2270 * bug fix: many threading/garbage collection symptoms sorted.
2271 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2272 might be pseudo-atomic.
2273 * internal change: Stopping for GC is now done with signals not
2274 ptrace. GC is now done in whichever thread wanted it, instead of
2276 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2278 * bug fix: result form in DO is not contained in the implicit
2280 * incompatible change: ICR structure is changed; the value part of
2281 CONTINUATION is now called LVAR; corresponding functions are
2282 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2283 * added type deriver for ISQRT (thanks to Robert E. Brown).
2284 * bug fix: better support for loading from the command line when an
2285 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2287 * fixed some bugs revealed by Paul Dietz' test suite:
2288 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2289 ** ROUND and FROUND now give the right answer when given very
2290 small float arguments.
2291 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2293 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2294 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2295 ** LCM should return a non-negative integer.
2296 ** PARSE-INTEGER returned the index of a terminator instead of the
2297 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2298 ** PARSE-INTEGER returned an incorrect index being applied to a
2300 ** LCM with two arguments of 0 returns 0 rather than signalling
2302 ** unsigned addition of a 32-bit constant with the high bit set no
2303 longer causes an internal compiler error on the x86.
2304 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2305 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2306 with negative last argument.
2307 ** byte specifiers with zero size and position no longer cause
2308 an error during type derivation.
2309 ** bignum multiplication on the Alpha platform now returns the
2311 * porting: The system now builds on SuSE AMD64, although it still
2312 generates a 32-bit binary.
2313 * .fasl file incompatibility: The fasl file version number has
2314 been incremented (because of the changes to internal compiler
2315 data structures referred to above).
2317 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2318 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2319 more accurately, on the Darwin kernel running on PowerPC hardware.
2320 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2321 * Compiler code deletion notes now signal a condition of type
2322 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2323 an associated MUFFLE-WARNING restart.
2324 * The compiler now performs limited argument count validation of
2325 constant format strings in FORMAT, and where appropriate in ERROR,
2326 CERROR and WARN. (thanks to Gerd Moellmann)
2327 * New ASDF-INSTALL contrib can be used for automatic download and
2328 installation of third-party Lisp code from CCLAN or other sites
2330 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2331 which forces another thread to execute a function supplied by the
2333 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2334 now accept and act upon their :ELEMENT-TYPE keyword argument.
2335 (reported by Martin Atzmueller, Edi Weitz)
2336 * bug fix: FILE-POSITION now accepts position designators up to
2337 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2338 is the greater. (thanks to Patrik Nordebo)
2339 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2340 Kalle Olavi Niemitalo)
2341 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2342 printable. (reported by Eric Marsden)
2343 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2344 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2345 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2346 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2347 (thanks to Stig Sandoe)
2348 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2349 UNREAD-CHAR now correctly includes the unread character in the
2350 target sequence. (thanks to Gerd Moellmann)
2351 * bug fix in threaded builds: the system can now be suspended and
2352 resumed by shell job control with minimal disruption.
2353 * bug fixes in times and timezones >2038AD
2354 * better handling of "where is GNU make?" problem in build scripts
2355 (thanks to Nikodemus Siivola)
2356 * new optimization: inside a named function any reference to a
2357 function with the same name is considered to be a self-reference;
2358 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2359 optimization quality.
2360 * new optimization on x86: logical functions and + now have
2361 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2362 used when the result is truncated to 32 bits.
2363 * VALUES declaration is partially enabled.
2364 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2365 * fixed some bugs revealed by Paul Dietz' test suite:
2366 ** The system now obeys the constraint imposed by
2367 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2368 types form a lattice under type intersection.
2369 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2370 ** ASSOC now ignores NIL elements in an alist.
2371 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2372 and (1+ MOST-POSITIVE-FIXNUM) answers.
2373 ** The addition of a method with invalid qualifiers to a generic
2374 function does not cause an error to be signalled immediately;
2375 a warning is signalled, and the error is generated only on
2376 calling the generic function.
2377 * changed .fasl file version number, in part to add type codes for
2378 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2379 obscure ANSI requirements
2381 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2382 * fixed bug 148: failure to inline-expand a local function left
2383 garbage, confusing the compiler.
2384 * fixed bugs 3cd: structure slot readers perform type check if the
2385 slot can have an invalid value (i.e. it is either not initialized
2386 or can be written with a less specific slot writer).
2387 * bug fix: the compiler now traps array references to elements off
2388 the end of an array; previously, the bounds checking in some
2389 circumstances could go off-by-one.
2390 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2392 * type declarations for array element types now obey the description
2393 on the CLHS page "Declaration TYPE", as per discussions on
2394 sbcl-help around 2003-05-08. This means that a declaration
2395 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2396 declaration, all references to BAR will be asserted or assumed
2397 (with THE, so dependent on compiler policy) to involve objects of
2398 type FOO. Note that no such declaration is implied in
2399 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2400 * declared types of functions from the "Conditions"
2401 chapter. (reported by Paul Dietz)
2402 * bug fix: CERROR accepts a function as its first argument.
2403 * bug fix: NTH an NTHCDR accept a bignum as index
2404 arguments. (reported by Adam Warner)
2405 * optimization: character compare routines now optimize comparing
2406 against a constant character. (reported by Gilbert Baumann)
2407 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2408 argument now works properly on the MIPS platform.
2409 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2410 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2411 declared to have a functional type, does not cause an error
2413 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2414 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2415 * fixed bug: if last continuation of a deleted block has a
2416 destination, this destination should be deleted too. (reported by
2418 * fixed a bug in the bootstrap process: the host compiler's values
2419 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2420 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2421 test case from Patrik Nordebo)
2422 * improved the ability of the disassembler on the PPC platform to
2423 provide helpful disassembly notes.
2424 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2425 the class in more cases than previously.
2426 * bug fix: FILE-POSITION now understands :START and :END for
2427 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2428 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2429 8) data. (thanks to David Lichteblau)
2430 * bug fix: it is possible to add a method to a generic function
2431 without lambda list.
2432 * bug fix: reader failed to signal END-OF-FILE inside an
2433 object representation. (reported by Nikodemus Siivola)
2434 * fixed some bugs revealed by Paul Dietz' test suite:
2435 ** LAST and [N]BUTLAST should accept a bignum.
2436 ** condition slot accessors are methods.
2437 ** (VECTOR NIL) is a subtype of STRING.
2439 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2440 * minor incompatible change: some nonsensical specialized lambda
2441 lists (used in DEFMETHOD) which were previously ignored now signal
2443 * minor incompatible change: the system is now aware of the types of
2444 variables in the COMMON-LISP package, and will signal errors for
2445 most violations of these type constraints (where previously they
2446 were silently accepted).
2447 * minor incompatible change: COMPILE-FILE now uses the freedom
2448 afforded (ANSI 3.2.2.3) to use derived function types for
2449 functions defined in the same file. This also permits the system
2450 to warn on static type mismatches and function
2451 redefinition. (Currently it does not work with high DEBUG level.)
2452 * minor incompatible change: VALUES declaration is disabled.
2453 * When issuing notes, the compiler now signals a condition of type
2454 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2455 restart for use in user handlers. It is expected that the
2456 COMPILER-NOTE condition will eventually become a condition
2457 supertype to a hierarchy of note types, which will then be
2458 handleable in a similar fashion. However, at the moment, no such
2459 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2460 but it's an implementation detail, not a classification for the
2462 * Changes in type checking closed the following bugs:
2463 ** type checking of unused values (192b, 194d, 203);
2464 ** template selection based on unsafe type assertions (192c, 236);
2465 ** type checking in branches (194bc).
2466 * A short form of VALUES type specifier has ANSI meaning (it has
2467 increased the number of situations when SBCL cannot perform type
2469 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2470 keywords or constants is permissible.
2471 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2472 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2473 classes. (thanks to Antonio Martinez)
2474 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2475 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2476 argument to a pretty-print stream. Instead, it leaves any such
2477 operations to DESCRIBE-OBJECT methods.
2478 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2480 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2481 lambda lists. (thanks to David Lichteblau)
2482 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2483 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2484 (thanks to Gerd Moellmann)
2485 * fixed bug 246: increased compilation speed of long
2486 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2488 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2490 * DEFCONSTANT now throws a condition of type
2491 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2492 constant to a non-EQL value; CONTINUE and ABORT restarts
2493 respectively change and preserve the value.
2494 * fixed bug 63: The code walker, part of the implementation of CLOS,
2495 is now better at handling symbol macros.
2496 * bug fix: There is no longer an internal implementation type named
2497 CL:LENGTH. (reported by Raymond Toy)
2498 * bug fix: In macro-like defining macros/special operators the
2499 implicit block does not enclose the lambda list.
2500 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2501 enclosed in parentheses) are not suitable as type specifiers, and
2502 their use properly signals an error now.
2503 * bug fix: An argument count mismatch for a type specifier in code
2504 being compiled no longer causes an unhandled error at compile
2505 time, but signals a compile-time warning.
2506 * fixed simple vector readable printing
2507 * bug fix: DESCRIBE takes more care over whether the class
2508 precedence list slot of a class is bound before accessing it.
2509 (reported by Markus Krummenacker)
2510 * bug fix: FORMATTER can successfully compile pretty-printer format
2511 strings which use variants of the ~* directive inside.
2512 * bug fix: SEARCH now applies its TEST predicate to the elements of
2513 the arguments in the correct order. (thanks to Wolfhard Buss)
2514 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2515 in type inference. (thanks to Robert E. Brown)
2516 * bug fix: Reading in symbols with an explicit package name of ""
2517 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2518 * fixed some bugs revealed by Paul Dietz' test suite:
2519 ** NIL is now allowed as a structure slot name.
2520 ** Arbitrary numbers, not just REALs, are allowed in certain
2521 circumstances in LOOP for-as-arithmetic clauses.
2522 ** Multiple class redefinitions before slot access no longer
2523 causes a type error.
2524 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2525 association between the name and a class.
2526 ** Generic functions with non-standard method-combination and over
2527 five methods all of which return constants no longer return NIL
2528 after the first few invocations. (thanks to Gerd Moellmann)
2529 ** CALL-NEXT-METHOD with no arguments now passes the original
2530 values of the arguments, even in the presence of assignment.
2531 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2533 ** DISASSEMBLE works with closures and funcallable instances.
2534 ** ADD-METHOD now returns the generic function, not the new method.
2535 ** FIND-METHOD signals an error if the lengths of the specializers
2536 is incompatible with the generic function, even if the ERRORP
2538 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2539 which its argument is a member.
2540 ** DEFCLASS only redefines the class named by its class-name
2541 argument if that name is the proper name of the class;
2542 otherwise, it creates a new class.
2543 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2544 of the UNBOUND-SLOT condition to the name of the slot.
2545 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2546 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2547 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2548 SLOT-MAKUNBOUND in the specified fashion.
2550 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2551 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2552 cross-compilation host. As a consequence, we can now bootstrap our
2553 way up to SBCL starting with a bare gcc toolchain and human-readable
2554 source code (first the source to CLISP, then the source to SBCL).
2555 * A contributed module containing a partial implementation of the
2556 simple-streams interface has been included. (thanks to Rudi
2558 * A contributed module implementing the RFC1321 Message Digest
2559 Algorithm, known as MD5, has been included.
2560 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2561 no longer has any effect, as the code controlled by this feature
2562 has been deleted. (As far as we know, no-one has ever built using
2563 this feature, and its semantics were confused in any case).
2564 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2565 work on conditions (as required by the ANSI specification),
2566 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2567 expected behaviour on conditions. Users should note, however,
2568 that such behaviour is not required by the ANSI specification,
2569 and so use of this behaviour may render their code unportable.
2570 * fixed some bugs revealed by Paul Dietz' test suite:
2571 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2573 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2574 ** SXHASH on condition objects no longer returns NIL.
2575 ** :ALLOCATION :CLASS slots are better treated; their values are
2576 updated on class redefinition, and initforms inherited from
2577 superclasses are applied.
2578 ** REMOVE-METHOD returns its generic function argument even when
2579 no method was removed.
2580 ** SHARED-INITIALIZE now initializes the values of the requested
2581 slots, including those with :ALLOCATION :CLASS.
2582 ** ALLOCATE-INSTANCE now works on structure classes defined via
2583 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2585 ** SLOT-EXISTS-P now works on conditions, as well as structures
2587 ** MAKE-LOAD-FORM now has the required methods on
2588 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2589 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2590 keyword, and now implements the SLOT-NAMES argument.
2591 ** methods with &OPTIONAL arguments no longer allow too many
2592 arguments to be passed in the call without error.
2593 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2594 option is consistent with the required arguments of the generic
2595 function lambda list.
2596 * bug fix: REQUIRE accepts a string designator. (Thanks to
2598 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2599 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2600 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2601 * bug fix: The compiler now checks for duplicated variables in macro
2603 * bug fix: SETQ on globals returns the correct value.
2604 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2605 not a COMPILER-ERROR (followed by some other strange error on
2606 choosing the CONTINUE restart).
2607 * bug fix: make.sh and friends are now more consistent in the way that
2608 they look for GNU "make".
2610 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2611 * experimental native threads support (on x86 Linux >=2.4 only).
2612 This is not compiled in by default: you need to add :SB-THREAD to
2613 the target features. See the "Beyond ANSI" chapter of the manual
2615 * fix for longstanding nonANSIism: The old distinction between
2616 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2617 The return value from CL:FIND-CLASS is now a CLOS class, and
2618 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2619 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2620 * An interface to the MetaObject Protocol, as described in Kiczales,
2621 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2622 MIT Press, 1991, is available from the SB-MOP package.
2623 * incompatible change: the SB-PCL package should now be considered
2624 a private implementation detail, and no longer a semi-private MOP
2626 * minor incompatible change: due to rearrangement for threads, the
2627 control stack and binding stack are now allocated at arbitrary
2628 addresses instead of being hardcoded per-port. Users affected by
2629 this probably have to be doing advanced things with shared
2630 libraries, and will know who they are.
2631 * minor incompatible change: Previously, all --eval forms used were
2632 processed with READ before any of them were processed with EVAL.
2633 Now each --eval form is processed with both READ and EVAL before
2634 the next --eval form is processed. (Thus package operations like
2635 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2636 work as the user might reasonably expect.)
2637 * minor incompatible change: *STANDARD-INPUT* is now only an
2638 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2640 * minor incompatible change: Y-OR-N-P is now character-oriented,
2641 not line oriented. Also, YES-OR-NO-P now works without errors.
2642 (thanks to Antonio Martinez)
2643 * sb-aclrepl module improvements: an integrated inspector, added
2644 repl features, and a bug fix to :trace command.
2645 * Known functions, which cannot be open coded by the backend, are
2646 considered to be able to check types of their arguments. (fixing
2647 a bug report by Nathan J. Froyd)
2648 * fixed a bug in computing method discriminating functions: It is
2649 now possible to define methods specialized on classes which have
2650 forward-referenced superclasses. (thanks to Gerd Moellmann)
2651 * fixed evaluation order in optional entries (reported by Gilbert
2653 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2654 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2655 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2656 required-by-AMOP NAME argument, as well as CLASS and
2657 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2658 * fixed bug 20: DEFMETHOD can define methods using names that are
2659 not the proper names of classes to designate class specializers.
2660 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2661 Unix file descriptors, instead of blowing up. (thanks to Antonio
2663 * Garbage collection refactoring: user-visible change is that a
2664 call to the GC function during WITHOUT-GCING will not do garbage
2665 collection until the end of the WITHOUT-GCING. If you were doing
2666 this you were probably losing anyway.
2667 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2668 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2669 TYPEP the latter but not the former.
2670 * The compiler issues a full WARNING for calls to undefined functions
2671 with names from the CL package.
2672 * MAP-INTO for a vector destination is open coded. (reported by
2673 Brian Downing on c.l.l)
2674 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2675 documentation string.
2676 * fixed some bugs revealed by Paul Dietz' test suite:
2677 ** COPY-ALIST now signals an error if its argument is a dotted
2679 ** Condition slots are now accessed more correctly in the presence
2680 of multiple initargs for a given slot.
2681 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2682 exclude restarts of the same name associated with a different
2684 ** DEFCLASS of forward-referenced classes with another
2685 forward-referenced class in the superclasses list no longer
2687 ** Condition slots are now initialized once each, not multiple
2688 times. (thanks to Gerd Moellmann)
2689 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2691 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2692 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2693 arguments contain duplicated elements.
2694 ** RESTART-CASE understands local macros.
2695 ** RESTART-CASE associates exactly its own restarts with a condition.
2696 ** ENDP in safe mode checks its argument to be of type LIST.
2697 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2698 in question is unbound.
2699 ** Optimized MAKE-INSTANCE functions no longer cause internal
2700 assertion failures in the presence of duplicate initargs.
2701 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2702 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2703 propagated even in the ctor optimized implementation of
2705 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2707 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2708 * a better implementation of SXHASH on (simple) bit vectors,
2709 measured both in execution speed and in distribution of results
2710 over the positive fixnums, has been installed. Likewise, a better
2711 implementation of EQUAL for simple bit vectors is now available.
2712 * fixed CEILING optimization for a divisor of form 2^k.
2713 * fixed bug 240 (emitting extra style warnings "using the lexical
2714 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2715 by Antonio Martinez)
2716 * fixed SXHASH, giving different results for NIL depending on type
2717 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2718 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2719 variable. (found by Rolf Wester)
2720 * SBCL does not ignore type declarations for special
2721 variables. (reported by rif on c.l.l 2003-03-05)
2722 * some bug fixes in contrib/sb-aclrepl/
2723 * fixed some bugs revealed by Paul Dietz' test suite:
2724 ** a bug in the CONS type specifier, whereby the CAR and CDR
2725 types got intertwined, has been fixed;
2726 ** the type system is now able to reason about the interaction
2727 between INTEGER and RATIO types more completely;
2728 ** APPEND, [N]REVERSE and NRECONC check that those their
2729 arguments, which must be proper lists, are really so;
2730 ** An array specialized to be unable to hold elements has been
2731 implemented, as required -- yes, really -- by ANSI;
2732 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2733 on malformed property lists;
2735 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2736 * incompatible packaging change: in line with Unix convention,
2737 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2738 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2739 * REQUIRE and PROVIDE are now optionally capable of doing something
2740 useful. See the documentation string for REQUIRE.
2741 * infrastructure for a managed SBCL contrib system: contributed
2742 modules in this release include:
2743 ** the ASDF system definition facility;
2744 ** an interface to the BSD Sockets API;
2745 ** an ACL-like convenience interface to the repl;
2746 (thanks to Kevin Rosenberg)
2747 ** an implementation of ROTATE-BYTE, with efficient implementation
2749 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2750 gives a closer approximation to the right answer than previously.
2751 (thanks to Raymond Toy)
2752 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2753 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2754 optional environment arguments, as required by ANSI.
2755 * fixed bugs in other functions taking environment objects, allowing
2756 calls with an explicit NIL environment argument to be compiled
2758 * fixed bug 228: primary return values from
2759 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2760 COMPILE or FUNCTION.
2761 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2762 :TYPE will no longer signal errors on innocuous objects.
2763 * fixed bug 231b: SETQ is better at respecting type declarations in
2764 the lexical environment.
2765 * fixed a bug in DEFCLASS: classes named by symbols with no or
2766 unprintable packages can now be defined.
2767 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2768 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2769 * fixed compiler failure related to checking types of functions.
2770 (reported by Robert E. Brown)
2771 * the compiler is now much more consistent in its error-checking
2772 treatment of bounding index arguments to sequence functions: in
2773 (SAFETY 3) code, errors will be signalled in almost all cases if
2774 invalid sequence bounding indices are passed to functions defined
2775 by ANSI to operate on sequences.
2776 * fixed a bug in the build procedure: documentation of SBCL-specific
2777 packages is now preserved and available in the final Lisp image.
2778 * lifted FDEFINITION lookup out of loops in the implementation of
2779 many list operations. (thanks to Robert E. Brown)
2780 * fixed a bug in the reader: the #n# reader macro now works for
2781 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2782 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2783 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2784 E. Brown and Tony Martinez respectively)
2785 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2786 SLOT-BOUNDP: in particular, it is now able to optimize them much
2787 better, and is now not vulnerable to having packages renamed.
2788 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2789 always called when a slot is not present in an instance. (thanks
2791 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2792 longer signal an error on unreading a character following EOF on
2793 the previous constituent stream. (thanks to Tony Martinez)
2794 * fixed some bugs revealed by Paul Dietz' test suite:
2795 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2796 not just nonnegative fixnums;
2797 ** the logical bit-array operators such as BIT-AND now accept an
2798 explicit NIL for their "opt-arg" argument (to indicate a
2799 freshly-consed result bit-array);
2800 ** ELT now signals an error on an invalid sequence index in safe
2802 ** the type system is now cleverer about negations of numeric
2803 types, and consequently understands the BIGNUM and RATIO types
2805 ** the type system is now cleverer about the interaction between
2806 INTEGER and RATIO types: while bugs still remain, many more
2807 cases are accurately computed;
2808 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2809 if it is in the last clause;
2810 ** CONSTANTLY now correctly returns a side-effect-free function in
2812 ** DECLARE is no longer treated as a special-operator; in
2813 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2814 * incremented fasl file version number due to the change in the
2815 DEFSTRUCT-SLOT-DESCRIPTION structure.
2817 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2818 * minor incompatible change: code processed by the "interpreter" or
2819 EVAL now has a compilation optimization policy of (DEBUG 2)
2820 (changed from (DEBUG 1)) to improve debuggability of interactive
2821 development, and to allow the use of the debug RETURN command in
2823 * an experimental implementation of the RETURN command for the
2824 debugger has been included. (thanks to Frederik Kuivinen)
2825 * fixed bug 62: constraints were not propagated into a loop.
2826 * fixed bug in embedded calls of SORT (reported and investigated by
2828 * fixed some bugs revealed by Paul Dietz' test suite:
2829 ** printing and reading of arrays with some dimensions having
2830 length 0 (thanks to Gerd Moellmann);
2831 ** BOA constructor with &AUX argument without a default value does
2832 not cause a type error;
2833 ** CONSTANTP now returns true for all self-evaluating objects.
2835 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2836 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2837 accessors that are related by inheritance, as specified in the
2838 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2839 Valtteri Vuorikoski)
2840 * The compiler is now able to inline functions that were defined in
2841 a complex lexical environment (e.g. inside a MACROLET).
2842 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2844 * Support for the upcoming FreeBSD-5.0 release has been included.
2845 (thanks to Dag-Erling Smorgrav)
2846 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2847 effect when it is not in a toplevel context.
2848 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2849 stand a better chance of being correct. (thanks to Gerd
2851 * fixed bug in COERCE, which now signals an error on coercing a
2852 rational to a bounded real type which excludes the expected
2854 * The compiler is now able to derive types more accurately from the
2855 COERCE and COMPILE functions.
2856 * fixed bug 223: functional binding is considered to be constant
2857 only for symbols in the CL package.
2858 * fixed bug 231: SETQ did not check the type of a variable being set
2859 (reported by Robert E. Brown)
2860 * A new optimization for MAKE-INSTANCE has been included, fixing
2861 various bugs (including relating to :ALLOCATION :CLASS slots and
2862 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2864 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2865 ** As required by ANSI, LOOP now disallows anonymous collection
2866 clauses such as COLLECT I in conjunction with aggregate boolean
2867 clauses such as THEREIS (= I 1);
2868 ** LOOP now signals an error when any variable is reused in the
2869 same loop (including the potentially useful construct analogous
2870 to WITH A = 1 WITH A = (1+ A);
2871 ** IT is only a special loop symbol within the first clause of a
2872 conditional loop clause;
2873 ** LOOP with a typed iteration variable over a hashtable now
2874 signals a type error iff it should.
2875 * fixed some other bugs revealed by Paul Dietz' test suite:
2876 ** FILE-STREAM now names the class previously known as FD-STREAM;
2877 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2878 argument) no longer signals an error;
2879 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2880 of the slot symbol, rather than using the current package
2881 ((:CONC-NAME "") continues to intern the slot's name in the
2883 * incremented fasl file version number, because of the incompatible
2884 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2885 of the new implementation of DEFINE-COMPILER-MACRO.
2887 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2888 * Support for building SBCL for MIPS platforms running in
2889 little-endian mode has now been checked in, and basic
2890 functionality on said platforms verified.
2891 * minor incompatible change: PCL now records the pathname of a file
2892 in which methods and the like are defined, rather than its
2894 * minor incompatible change: TRUENAME now considers the truename of
2895 a file naming a directory to be the pathname with :DIRECTORY
2896 component indicating that directory.
2897 * minor incompatible change: a NAMED clause in the extended form of
2898 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2899 reason for the previous behaviour is unclear.
2900 * more systematization and improvement of CLOS and MOP conformance
2901 in PCL (thanks to Gerd Moellman and Pierre Mai):
2902 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2904 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2905 primary methods with no specializers;
2906 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2908 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2909 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2910 FORWARD-REFERENCED-CLASSes; error reporting on
2911 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2913 ** SXHASH on CLOS instances now uses a slot internal to the
2914 instance to return different numbers on distinct instances,
2915 while preserving the same return value through invocations of
2917 ** DEFMETHOD signals errors when methods with longer incongruent
2918 lambda lists are added to generic functions;
2919 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2920 CLASS, as specified in AMOP;
2921 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2922 on the order returned by the primary method for classes of
2923 class STANDARD-CLASS;
2924 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2925 * fixed some bugs shown by Paul Dietz' test suite:
2926 ** DOLIST puts its body in TAGBODY;
2927 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2929 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2930 value producing form;
2931 ** if more variables are given to PROGV than values, extra
2932 variables are bound and made to have no value;
2933 ** NSUBSTITUTE on list arguments gets the right answer with
2935 ** ELT signals an error of type TYPE-ERROR when the index argument
2936 is not a valid sequence index;
2937 ** LOOP signals (at macroexpansion time) an error of type
2938 PROGRAM-ERROR when duplicate variable names are found;
2939 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2940 ** FUNCALL of special-operators now cause an error of type
2942 ** PSETQ now works as required in the presence of side-effecting
2943 symbol-macro places;
2944 ** NCONC accepts any object as its last argument;
2945 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2947 ** loop-for-as-package does not require a package to be explicitely
2949 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2951 * fixed bug 166: compiler preserves "there is a way to go"
2952 invariant when deleting code.
2953 * fixed bug 172: macro lambda lists with required arguments after
2954 &REST arguments now cause an error to be signalled. (thanks to
2956 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2957 bug. (thanks to Matthew Danish)
2958 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2960 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2961 arguments are lexically rebound. (thanks to Gerd Moellmann and
2963 * fixed bug 194: error messages are now more informative when there
2964 is no primary method applicable in a call to a generic
2965 function. (thanks to Gerd Moellmann)
2966 * fixed bug in command line argument checking (thanks to Julian
2968 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2969 (thanks to Matthew Danish)
2970 * incremented fasl file version number, because of the
2971 SXHASH-related changes in the layout of CLOS data structures
2973 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2974 * minor incompatible change: The runtime (the Unix executable named
2975 "sbcl") is now much pickier about the .core files it will load.
2976 Essentially it now requires .core files to descend from the same
2977 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2978 as the runtime does. (The intent is to prevent the crashes which
2979 can occur, and which can even be reported as mysterious failures,
2980 when people patch the sources or change the build parameters
2981 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2982 sbcl and .core files.)
2983 * fixed bug: VALUES-LIST is no longer optimized away.
2984 * fixed bug 142: The FFI conversion of C string values to Lisp
2985 string values no longer conses excessively. (thanks to Nathan
2986 Froyd porting Raymond Toy's fix to CMU CL)
2987 * began to systematize and improve MOP conformance in PCL (thanks to
2988 Nathan Froyd, Gerd Moellman and Pierre Mai):
2989 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2991 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2992 ** FINALIZE-INHERITANCE is now called on class finalization;
2993 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2994 argument precedence order.
2995 * fixed bug 202: The compiler no longer fails on functions whose
2996 derived types contradict their declared type.
2997 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2998 so it can be non-toplevel.
2999 * The fasl file version number has changed (because of the new
3000 implementation of DEFMACRO).
3001 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3002 safe code, that any length requirement by their type-specifier
3003 argument is valid. The exceptions to this are described in bug
3005 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3006 functions now check the types of their inputs as required by ANSI.
3007 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3008 introduced symbol is DECLAREd to be SPECIAL.
3009 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3010 * fixed bug: :COUNT argument to sequence functions may be negative
3011 * fixed bug: body of DO-SYMBOLS may contain declarations
3012 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3013 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3014 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3015 symbol macro only once
3016 * fixed printing of call frame when argument list is unavailable
3017 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3018 * compiler no longer signals WARNING on unknown keyword
3021 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3022 * A beta-quality port to the mips architecture running Linux,
3023 based on the old CMUCL backend, has been made. It has been tested
3024 on a big-endian kernel, and works sufficiently well to be able to
3025 rebuild itself; it has not been tested in little-endian mode.
3026 * fixed an inconsistency between gencgc.c and purify.c which made
3027 dumping/loading .core files unreliable
3028 * fixed bug 120a: The compiler now deals correctly with IFs where
3029 the consequent is the same as the alternative, instead of
3030 misderiving the return type. (thanks to Alexey Dejneka)
3031 * fixed bug 113: Logical pathnames are now dumpable (the logical
3032 host is resolved at load-time, throwing an error if it is not
3034 * fixed bug 174: FORMAT's error message is slightly clearer when a
3035 non-printing character is used in a format directive.
3036 * fixed several bugs in compiler checking of type declarations, i.e.
3037 violations of the Python "declarations are assertions" principle
3038 (thanks to Alexey Dejneka)
3039 * fixed several bugs in PCL's error checking (thanks to Gerd
3041 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3043 * fixed bug in compilation of functions as first class values
3044 (thanks to Antonio Martinez-Shotton)
3045 * The compiler's handling TYPE-ERRORs which it can prove will
3046 inevitably happen at runtime has been cleaned up and corrected
3047 in several ways. (thanks to Alexey Dejneka)
3048 * improved argument type checking for various basic arithmetic
3049 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3050 lavished on them in the past that they can be compiled in many
3051 ways in different special cases
3052 * fixed bug 181: compiler checks validity of user supplied type
3054 * cleaned up code flushing in optimization: Function calls which
3055 should signal errors for safety purposes (e.g. which ANSI says
3056 should signal errors when their arguments are of incorrect type)
3057 are no longer optimized away.
3058 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3059 * incremented fasl file version number, because changes in the
3060 implementation of sequence functions like COERCE caused
3061 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3062 in old inline expansions) to become undefined. (Actually these
3063 changes were later undone, so we might very well be binary
3064 compatible with 0.7.7 after all, but leaving the version number
3065 incremented seemed like the simplest and most conservative
3068 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3069 * An alpha-quality port to the parisc architecture running Linux,
3070 based on the old CMUCL backend, has been made. This, even more so
3071 than the other backends, should be considered still a work in
3072 progress; known problems include that the Linux kernel in 64-bit
3073 mode does not propagate the correct sigcontext structure to
3074 userspace, and consequently SBCL on a parisc64 kernel will not
3076 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3077 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3078 suppressed.) Also now that the compiler is looking at declarations
3079 in the environment, it checks optimization declarations as well,
3080 and suppresses inlining when (> DEBUG SPEED).
3081 * More fixes have been made to treatment of floating point exception
3082 treatment and other Unix signals. In particular, floating point
3083 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3084 * The detection and handling of control stack exhaustion (infinite
3085 or very deeply nested recursion) has changed. Stack exhaustion
3086 detection is now done by write-protecting pages at the OS level
3087 and applies at all optimization settings; when found, a
3088 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3089 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3090 be caught using IGNORE-ERRORS.
3091 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3092 correctly when transforming recursive LABELS functions to LETs.
3093 (thanks to Alexey Dejneka)
3094 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3095 that are names of constants or global variables.
3096 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3097 alien routines with docstrings.
3098 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3099 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3101 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3102 (thanks to Eric Marsden)
3103 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3104 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3105 * Bug fix: The SPARC backend can now compile functions involving
3106 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3107 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3108 to a functions expecting a PATHNAME-DESIGNATOR.
3109 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3110 lambda lists. (thanks to Alexey Dejneka)
3111 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3112 functions so that EXPORTing the name of the class doesn't cause
3113 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3114 to Antonio Martinez for reporting this)
3115 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3116 constant-folded without failing an assertion. (thanks to Einar
3117 Floystad Dorum for reporting this)
3118 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3119 types are now dealt with more correctly by the compiler.
3120 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3121 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3122 behaviour analogous to RENAME-FILE. This puts its behaviour more
3123 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3125 * The fasl file version number has changed again. (because of the
3126 bug fix involving the names of PCL MAKE-INSTANCE functions)
3128 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3129 * bug fix: Floating point exceptions are treated much more
3130 consistently on the x86/Linux and PPC/Linux platforms.
3131 * Array initialization with :INITIAL-ELEMENT is now much faster for
3132 cases when the compiler cannot open code the array creation, but
3133 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3134 array accesses have also seen a speed increase.
3135 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3136 specified. (This was at the root of some bad interactions between
3137 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3139 * bug fix: Internal error arguments for undefined functions are now
3140 computed correctly on the PPC/Linux platform.
3141 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3142 Raymond Toy's patch for CMU CL)
3143 * Support for the Solaris 9 operating environment has been included
3144 (thanks to Daniel Merritt)
3145 * A very ugly but hopefully complete draft of the missing FFI chapter
3146 of the manual has been created by reformatting the corresponding
3147 CMU CL manual chapter into (currently very ugly and incoherent)
3148 DocBook and bringing it up to date for SBCL behavior. Thus, the
3149 manual is now essentially complete, at least by my extreme
3150 once-and-only-once standards, whereby it's acceptable to refer to
3151 the doc strings of SB-EXT functions as the primary documentation.
3152 * The fasl file version number has changed again, due to cleanup of
3153 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3154 is no longer a static symbol.)
3156 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3157 * SBCL now builds with OpenMCL (version 0.12) as the
3158 cross-compilation host; also, more progress has been made toward
3159 bootstrapping under CLISP.
3160 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3162 * bug 158 fixed: The compiler can now deal with integer loop
3163 increments different from 1; fixing this turned out also to fix
3165 * bug 169 fixed: no more bogus warnings about using lexical bindings
3166 despite the presence of perfectly good SPECIAL declarations (thanks
3167 to David Lichteblau)
3168 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3169 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3170 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3172 * bug 180 fixed: Method combination specifications no longer ignore
3173 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3174 * bug fix: Structure type predicate functions now check their argument
3175 count as they should.
3176 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3177 correctly. (thanks to Pierre Mai)
3178 * minor incompatible change: The --noprogrammer option is deprecated
3179 in favor of the new --disable-debugger option, which is very similar.
3180 (The major difference is that it takes effect at a slightly different
3181 time at startup, causing handling of errors in --sysinit and
3182 --userinit files will be affected differently.) The
3183 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3184 been added to allow this functionality to be controlled from ordinary
3185 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3186 maintainers, who might want to run non-interactive scripts to
3187 build SBCL cores which will later be used interactively.)
3188 * minor incompatible change: The LOAD function no longer, when given
3189 a wild pathname to load, loads all files matching that pathname.
3190 Instead, an error of type FILE-ERROR is signalled.
3192 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3193 * bug 147 fixed: The compiler preserves its block link/count
3194 invariants more correctly now so that it doesn't crash. (thanks
3196 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3198 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3199 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3200 no-value-for-this-character as NIL (as specified by ANSI).
3201 * HOST-NAMESTRING on physical pathnames now returns a string that is
3202 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3203 (thanks to Christophe Rhodes)
3204 * The Alpha port handles icache flushing more correctly. (thanks to
3206 * More progress has been made toward bootstrapping under CLISP. (thanks
3207 to Christophe Rhodes)
3208 * The fasl file format has changed again, because dynamic loading
3209 on OpenBSD (which has non-ELF object files) motivated some cleanups
3210 in the way that foreign symbols are transformed and passed around.
3211 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3212 is no longer treated as whitespace by the reader, but instead as
3213 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3214 |A
\7fB|, instead of A as it used to.
3216 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3217 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3218 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3219 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3220 this as of 0.7.1.45, but was left out of the previous news section
3221 (thanks to Dan Barlow)
3222 * SBCL now runs on the Solaris operating system on SPARC architectures
3223 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3224 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3225 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3226 * SPARC backend cleanups, allowing builds of cores optimized for V8
3227 and V9 SPARCS, and also emission of code targeted to a particular
3228 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3230 * SBCL is closer to bootstrapping under CLISP, thanks to various
3231 fixes by Christophe Rhodes.
3232 * The fasl file format has changed again, to allow the compiler's
3233 INFO database to support symbol macros.
3234 * The user manual (in doc/) is formatted into HTML more nicely.
3235 (thanks to coreythomas)
3236 * The system is smarter about SUBTYPEP relationships, especially
3237 those involving NOT types (including types such as ATOM which are
3238 represented internally using NOT types). Thus SUBTYPEP is less
3239 likely to return (VALUES NIL NIL) in general, and in particular
3240 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3242 * The fasl file format has changed again, because the internal
3243 representation of types now includes a new slot to support the new
3244 SUBTYPEP-of-NOT-types logic.
3245 * (not a change in the main branch of SBCL, but a related prototype
3246 which can hopefully be merged into the main branch of SBCL in the
3247 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3248 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3249 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3250 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3251 dispatch macro character combinations. (thanks to Alexey Dejneka)
3252 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3253 that superficially look like logical namestrings correctly.
3254 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3256 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3257 default supplied-p) for &optional and &key arguments. (thanks to
3260 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3261 * incompatible change: The compiler is now less aggressive about
3262 tail call optimization, doing it only when (> SPACE DEBUG) or
3263 (> SPEED DEBUG). (This is an incompatible change because there are
3264 programs which relied on the old CMU-CL-style behavior to optimize
3265 away their unbounded recursion which will now die of stack overflow.)
3266 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3267 for non-GENCGC systems has been increased to 20M (since that
3268 seems much closer to the likely performance optimum for modern
3269 systems than the old 4M value was)
3270 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3271 and *DEBUG-PRINT-LEVEL*
3272 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3273 of CMU CL's support for SPARC, and various endianness and other
3274 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3275 * new syntactic sugar for the Unix command line: --load foo.bar is now
3276 an alternate notation for --eval '(load "foo.bar")'.
3278 ** The system now detects stack overflow and handles it gracefully,
3279 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3280 optimization settings. (This is a good thing in general, and
3281 its introduction in this version should be particularly timely
3282 for anyone whose code fails because of suppression of tail
3284 ** The system now hunts for the C variable "environ" in a more
3285 devious way, to avoid segfaults when the C library version differs
3286 between compile time and run time. (thanks to Christophe Rhodes)
3287 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3288 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3289 ** The compiler no longer issues bogus style warnings for undefined
3290 classes in the same source file as the DEFCLASSes which defined
3291 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3293 ** fixes in CONDITION class precedence list for undefined function
3294 errors (thanks to Alexei Dejneka)
3295 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3296 correctly. (thanks to Dan Barlow)
3297 ** portability fixes aiming at bootstrapping under CLISP (thanks
3298 to Dave McDonald and Christophe Rhodes)
3299 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3300 ** fixes in type translation and and type inference (thanks to
3302 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3303 ** various fixes in the new ports (thanks to Dan Barlow)
3304 * several changes related to debugging:
3305 ** suppression of tail recursion, as noted above
3306 ** stack overflow detection, as noted above
3307 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3308 is now the default. (For some time encapsulation has been more
3309 reliable than the breakpoint-based :ENCAPSULATE NIL
3310 implementation, at least on X86 systems; and I just noticed that
3311 encapsulation also seems closer to the spirit of the ANSI
3314 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3316 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3317 up properly again. (There was a packaging bug in 0.7.0 which
3318 left their definitions in SB-SYS::LOAD-FOREIGN and
3319 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3320 vital for most things which interface to C-level interfaces,
3321 like extensions working with sockets or databases or
3322 Perl-compatible regexes or whatever, and the need to fix
3323 this bug is the main reason that 0.7.1 was released so
3325 ** DEFGENERIC is now choosier about the methods it redefines, so that
3326 reLOADing a previously-LOADed file containing DEFGENERICs does
3327 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3328 cycle now works as it should. (thanks to Alexey Dejneka)
3329 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3330 fixed by Christophe Rhodes. (He actually submitted this patch
3331 months ago, and I delayed until after 0.7.0.)
3332 ** Bug 111 (internal compiler confusion about runtime checks on
3333 FUNCTION types) was fixed by Alexey Dejneka.
3334 * Some internal cleanups (getting rid of variables which aren't
3335 needed now that the byte interpreter is gone) caused the fasl
3336 file format number to change again.
3338 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3339 * major incompatible change: The default fasl file extension, i.e. the
3340 default extension for files produced by COMPILE-FILE, has changed
3341 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3343 ** There are many changes in the implementation of the compiler.
3344 SBCL is now essentially a compiler-only implementation of ANSI
3345 Common Lisp. EVAL still "interprets" a few special cases, but
3346 almost all the interesting cases are handled by creating
3347 a LAMBDA expression, calling COMPILE on it, then calling
3348 FUNCALL on the result.
3349 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3350 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3351 Since the code is newer, there might still be some new bugs
3352 (though not as many as before Martin Atzmueller's fixes:-). But
3353 the new code is substantially simpler and clearer, and hopefully
3354 any remaining bugs will be simpler, less fundamental, and more
3355 fixable then the bugs in the old code.
3356 ** The revised compiler is still a little unsteady on its feet.
3358 *** The debugging information it produces (particularly the names
3359 of FUNCTION objects) is sometimes much less useful than what
3360 the old compiler produced.
3361 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3362 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3363 within a MACROLET) has been temporarily weakened.
3364 ** There are new compiler optimizations for various functions:
3365 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3366 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3367 *** the math functions TRUNCATE, FLOOR, and CEILING
3368 *** the function-of-all-trades COERCE
3369 Mostly these should be transparent, but there's one
3370 potentially-annoying problem (bug 117): when the compiler
3371 inline-expands a function and does type analysis on the result,
3372 it can create control paths which have type mismatches, and
3373 when it can't prove that those control paths aren't taken,
3374 it will issue WARNINGs about the type mismatches. This is
3375 a particular problem in practice for the new sequence functions.
3376 It's not clear how this should be fixed, and for now, a
3377 workaround is given in the entry for 117 in the BUGS file.
3378 ** (Because of the interaction between the two previous items --
3379 occasional inlining problems and new inline expansions -- some
3380 of the new sequence function optimizations won't really kick in
3381 completely until debugging information, and then inlining, are
3382 straightened out in some future version.)
3383 * minor incompatible changes:
3384 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3385 DIRECTORY no longer implicitly promotes NIL slots of its
3386 pathname argument to :WILD. In particular, when you ask for the
3387 contents of a directory (which you used to be able to do without
3388 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3389 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3390 ** changes in behavior that ANSI explicitly defines to be
3391 implementation dependent:
3392 *** The new compiler-only implementation still conforms with ANSI,
3393 but acts a little different than before. Besides the obvious
3394 changes in performance tradeoffs (that the cost per form passed
3395 to EVAL has gone up, and the cost per form executed by EVAL
3396 has gone down), the behavior of the system changes a little
3397 because there are no longer any interpreted function objects.
3398 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3399 e.g. doing COMPILE on the output of interactive DEFUN is
3401 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3403 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3404 has changed from (:CL) to NIL.
3405 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3406 e.g. #\Newline and #\Tab, don't have names specified in the
3407 ANSI Common Lisp standard, is now based on their ASCII symbolic
3408 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3409 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3410 are no longer used for output.
3411 ** changes in internal implementation constants:
3412 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3413 4 million. (If your application spends a lot of time GCing and
3414 you have a lot of RAM, you might want to experiment with
3415 increasing it even more.)
3416 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3417 However, almost all old code should still continue to work without
3418 immediate update, as SB-C-CALL is now a (deprecated) nickname
3420 ** Old operator names in the style DEF-FOO are now deprecated in
3421 favor of new corresponding names DEFINE-FOO, for consistency with
3422 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3423 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3424 internal symbols, but a few supported extensions like
3425 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3426 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3427 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3428 etc. as you get deeper into recursive calls to the debugger
3429 command loop, instead of the old "5]", "5]]", "5]]]"
3430 sequence. (I was motivated to do this when squabbles between
3431 ILISP and SBCL left me very deeply nested in the debugger. In the
3432 short term, this change will probably provoke more ILISP/SBCL
3433 squabbles, but hopefully it will be an improvement in the long run.)
3434 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3435 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3436 function, since declaiming return types involving aliens is
3437 (1) annoyingly messy to do by hand and (2) vital to efficient
3438 compilation of code which calls such functions.
3439 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3440 longer reexported by the SB-EXT package. They're solely useful
3441 for alien code, so it seems more logical that you should get
3442 them from the SB-ALIEN package, not in SB-EXT.
3443 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3444 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3445 features. Instead, the code that they used to control is always
3446 built into the system.
3447 * many other bug fixes
3448 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3449 advantage of the new EVAL-WHEN stuff and to clean them up in
3450 general, and they are now more ANSI-compliant in a number of
3451 ways. Martin Atzmueller is responsible for a lot of this.
3452 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3454 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3455 *** correct ERROR type for various file operations
3456 *** some fixes for Lisp streams
3457 *** DEFMETHOD syntax checking
3458 *** changing old weird representation of debug information as
3459 strings (which, among their other deficiencies, don't transform
3460 correctly when you rename packages, and don't change their
3461 print representation when you change things like *PACKAGE*
3462 and *PRINT-LENGTH*) to symbols and lists of symbols
3463 He also made several improvements and fixed several bugs in DESCRIBE.
3464 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3466 *** misbehavior of WRITE-STRING/WRITE-LINE
3467 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3468 and several other LOOP problems as well
3469 *** DIRECTORY when similar filenames are present
3470 *** DEFGENERIC with :METHOD options
3471 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3472 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3473 *** argument ordering in FIND with :TEST option
3474 *** mishandled package designator argument in APROPOS-LIST
3475 *** various problems in the backquote readmacro
3476 *** a bug in APROPOS
3477 *** probably some others that I'm not describing very well here,
3478 since the CVS log documents them by reference to sbcl-devel
3479 messages, and the SourceForge archives aren't working well.:-(
3480 ** Dan Barlow improved the Alpha port (and is making progress on the
3481 PPC port, for those of you who think different).
3482 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3483 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3484 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3485 continues to work on the SPARC port (for those of you in a position
3486 to look down upon our little PC-compatible boxes from a great height).
3487 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3488 and uses that copy, rather than the current dynamic value, when
3489 it's trying to decide whether to truncate output. Thus e.g.
3490 (let ((*print-lines* 50))
3491 (pprint-logical-block (stream nil)
3493 (let ((*print-lines* 8))
3494 (print (aref possiblybigthings i) stream)))))
3495 should now truncate the logical block only at 50 lines, instead of
3496 often truncating it at 8 lines, as it did before.
3497 * The doc/cmucl/ directory, containing old CMU CL documentation
3498 from the time of the fork, is no longer part of the base system.
3499 SourceForge has shut down its anonymous FTP service, and with it
3500 my original plan for distributing the old CMU CL documentation
3501 there. For now, if you need these files you can download an old
3502 SBCL source release and extract them from it.
3503 * The fasl file version number changed again, for dozens of reasons,
3504 some of which are apparent above.
3506 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3507 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3508 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3509 patch, so that DISASSEMBLE output is much nicer.
3510 * The code in the SB-PROFILE package now seems reasonably stable.
3511 I still haven't decided what the final interface should look like
3512 (I'd like PROFILE to interact cleanly with TRACE, since both
3513 facilities use function encapsulation) but if you have a need
3514 for profiling now, you can probably use it successfully with
3515 the current CMU-CL-style interface.
3516 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3517 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3518 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3519 link now returns the dangling link itself, and for similar
3520 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3521 link itself. (In these cases the old code signalled an error and
3522 looped endlessly, respectively.) Thus, DIRECTORY now works even
3523 in the presence of dangling and cyclic symbolic links.
3524 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3525 is now a supported extension again, since the consensus on
3526 sbcl-devel was that it can be useful for ordinary development
3527 work, not just for debugging SBCL itself.
3528 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3529 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3530 that currently-defined functions might be redefined later with
3531 different return types.
3532 * Hash tables can be printed readably, as inspired by CMU CL code
3533 of Eric Marsden and SBCL code of Martin Atzmueller.
3534 * better error handling in CLOS method combination, thanks to
3535 Martin Atzmueller porting Pierre Mai's CMU CL patches
3536 * more overflow fixes for >16Mbyte I/O buffers
3537 * A bug in READ has been fixed, so that now a single Ctrl-D
3538 character suffices to cause end-of-file on character streams.
3539 In particular, now you only need one Ctrl-D at the command
3540 line (not two) to exit SBCL.
3541 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3543 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3544 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3545 does the right thing.
3546 * The implementation of some type tests, especially for CONDITION
3547 types, is now tidier and maybe faster, due to CMU CL code
3548 originally by Douglas Crosher, ported by Martin Atzmueller.
3549 * Some math functions have been fixed, and there are new
3550 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3551 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3552 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3553 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3554 likely to be built into sbcl-0.7.0.)
3555 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3556 CL, ported to SBCL by Martin Atzmueller)
3557 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3558 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3559 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3560 * There's a new slam.sh hack to shorten the edit/compile/debug
3561 cycle for low-level changes to SBCL itself, and a new
3562 :SB-AFTER-XC-CORE target feature to control the generation of
3563 the after-xc.core file needed by slam.sh.
3564 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3565 COMPILE-FILE is no longer supported, so that now every function
3566 gets an entry point, so that block compilation looks a little
3567 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3568 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3569 returns the number of bytes consed since the system started,
3570 rather than the number consed since the first time the function
3571 was called. (The new definition parallels ANSI functions like
3572 CL:GET-INTERNAL-RUN-TIME.)
3573 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3574 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3575 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3576 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3577 * Fasl file version numbers are now independent of the target CPU,
3578 since historically most system changes which required version
3579 number changes have affected all CPUs equally. Similarly,
3580 the byte fasl file version is now equal to the ordinary
3583 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3584 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3585 is no longer recognized. I apologize for this, because it was
3586 listed in SB-EXT as a supported extension, but I found that
3587 its existing behavior was poorly specified, as well as incorrectly
3588 specified, and it looked like too much of a mess to straighten it
3589 out. I have enough on my hands trying to get ANSI stuff to work..
3590 * many patches ported from CMU CL by Martin Atzmueller, with
3591 half a dozen bug fixes in pretty-printing and the debugger, and
3592 half a dozen others elsewhere
3593 * fixed bug 13: Floating point infinities are now supported again.
3594 They might still be a little bit flaky, but thanks to bug reports
3595 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3596 as flaky as they were.
3597 * The --noprogrammer command line option is now supported. (Its
3598 behavior is slightly different in detail from what the old man
3599 page claimed it would do, but it's still appropriate under the
3600 same circumstances that the man page talks about.)
3601 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3602 are now supported, and enabled by default. Thus, the compiler can
3603 handle many floating point and complex operations much less
3604 inefficiently. (Thus e.g. you can implement a complex FFT
3606 * The compiler now detects type mismatches between DECLAIM FTYPE
3607 and DEFUN better, and implements CHECK-TYPE more correctly, and
3608 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3609 thanks to patches from Martin Atzmueller.
3610 * various fixes to make the cross-compiler more portable to
3611 ANSI-conforming-but-different cross-compilation hosts (notably
3612 Lispworks for Windows, following bug reports from Arthur Lemmens)
3613 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3614 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3615 fixed thanks to Pierre Mai's CMU CL patches.
3616 * fixes to keep the system from overflowing internal counters when
3617 it tries to use i/o buffers larger than 16M bytes
3618 * fixed bug 45a: Various internal functions required to support
3619 complex special functions have been merged from CMU CL sources.
3620 (When I was first setting up SBCL, I misunderstood a compile-time
3621 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3622 * improved support for type intersection and union, fixing bug 12
3623 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3624 more obscure bugs as well
3625 * some steps toward byte-compiling non-performance-critical
3626 parts of the system, courtesy of patches from Martin Atzmueller
3627 * Christophe Rhodes has made some debian packages of sbcl at
3628 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3629 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3630 stable, but are nonetheless usable. When he's ready, I'd be happy
3631 to add them to the SourceForge "File Releases" section. (And if
3632 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3633 * new fasl file format version number (because of changes in
3634 internal representation of (OR ..) types to accommodate the new
3635 support for (AND ..) types, among other things)
3637 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3638 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3640 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3642 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3643 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3644 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3645 * DESCRIBE now works on structure objects again.
3646 * Most function call argument type mismatches are now handled as
3647 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3648 know whether the function will be redefined before the call is
3649 executed. (The compiler could flag local calls with full WARNINGs,
3650 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3651 it doesn't keep track of enough information to know whether calls
3652 are local in this sense.)
3653 * Compiler output is now more verbose, with messages truncated
3654 later than before. (There should be some supported way for users
3655 to override the default verbosity, but I haven't decided how to
3656 provide it yet, so this behavior is still controlled by the internal
3657 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3658 src/compiler/ir1util.lisp.)
3659 * Fasl file format version numbers have increased again, because
3660 support for the Gray streams extension changes the layout of the
3661 system's STREAM objects.
3662 * The Gray subclassable streams extension now works, thanks to a
3663 patch from Martin Atzmueller.
3664 * The full LOAD-FOREIGN extension (not just the primitive
3665 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3666 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3667 but like most other programs, it defaults to copying the Unix
3668 environment from the original process instead of starting the
3669 new process in an empty environment.
3670 * Extensions which manipulate the Unix environment now support
3671 an :ENVIRONMENT keyword option which doesn't smash case or
3672 do other bad things. The CMU-CL-style :ENV option is retained
3673 for porting convenience.
3674 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3675 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3677 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3679 * A patch from Martin Atzmueller seems to have solved the SIGINT
3680 problem, and as far as we know, signal-handling now works cleanly.
3681 (If you find any new bugs, please report them!)
3682 * The system no longer defaults Lisp source file names to types
3683 ".l", ".cl", or ".lsp", but only to ".lisp".
3684 * The compiler no longer uses special default file extensions for
3685 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3686 seems to expect a single default extension for all compiled code,
3687 and there's no compelling reason to try to stretch the standard
3688 to allow two different extensions.) Instead, byte-compiled files
3689 default to the same extension as native-compiled files.
3690 * Fasl file format version numbers have increased again, because
3691 a rearrangement of internal implementation packages made some
3692 dumped symbols in old fasl files unreadable in new cores.
3693 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3694 many fewer weird special cases.
3695 * Bug #17 (differing COMPILE-FILE behavior between logical and
3696 physical pathnames) has been fixed, and some related misbehavior too,
3697 thanks to a patch from Martin Atzmueller.
3698 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3699 by Tim Moore, ported to SBCL by Martin Atzmueller.
3700 * Martin Atzmueller fixed several filesystem-related problems,
3701 including bug #36, in part by porting CMU CL patches, which were
3702 written in part by Paul Werkowski.
3703 * More compiler warnings in src/runtime/ are gone, thanks to
3704 more patches from Martin Atzmueller.
3705 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3708 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3710 * DESCRIBE now works on CONDITION objects.
3711 * The debugger now handles errors which arise when trying to print
3712 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3714 * The build system now uses an additional file, customize-target-features.lisp,
3715 to allow local modifications to the target *FEATURES* list. (The point of
3716 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3717 debugging features enabled, without having to worry about propagating it
3718 into everyone's system when I do a "cvs update".) When no
3719 customize-target-features.lisp file exists, the target *FEATURES* list
3720 should be constructed the same way as before.
3721 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3722 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3723 warn about behavior which is undefined under the ANSI spec. Note: This
3724 is specified by ANSI, but it's not very popular with programmers.
3725 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3726 macro in the SBCL sources for an example of a workaround which you
3727 might use to make portable ANSI-standard code which does what you want.
3728 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3729 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3730 when it's used as a non-toplevel form. (This is required in order
3731 to implement the DEFCONSTANT-EQX macro.)
3732 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3733 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3734 funny things, due to bugs in the implementation of EVAL-WHEN
3735 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3736 non-toplevel DEFCONSTANTs are uncommon.)
3737 * The core file version number and fasl file version number have been
3738 incremented, because the old noncompliant DEFCONSTANT behavior involved
3739 calling functions which no longer exist, and because I also took the
3740 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3741 * fixed bug 1 (error handling before read-eval-print loop starts), and
3742 redid debugger restarts and related debugger commands somewhat while
3744 ** The QUIT debugger command is gone, since it did something
3745 rather different than the SB-EXT:QUIT command, and since it never
3746 worked properly outside the main toplevel read/eval/print loop.
3747 Invoking the new TOPLEVEL restart provides the same functionality.
3748 ** The GO debugger command is also gone, since you can just invoke
3749 the CONTINUE restart directly instead.
3750 ** The TOP debugger command is also gone, since it's redundant with the
3751 FRAME 0 command, and since it interfered with abbreviations for the
3753 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3755 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3756 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3757 some porting work by Martin Atzmueller)
3758 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3759 with a boatload of patches which he ported to SBCL. Now that those
3761 ** The system tries to make sure that its low-priority messages
3762 are prefixed by semicolons, to help people who like to use
3763 syntax highlighting in their ILISP buffer. (This patch
3764 was originally due to Raymond Toy.)
3765 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3766 patches originally written by Raymond Toy.
3767 ** The compiler understands coercion between single-value and
3768 multiple-VALUES type expressions better, getting rid of some very
3769 weird behavior, thanks to patches originally by Robert MacLachlan
3770 and Douglas Crosher.
3771 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3772 lambda lists, thanks to a patch originally by Pierre Mai.
3773 ** The system no longer bogusly warns about "abbreviated type
3775 ** The compiler gets less confused by inlining and RETURN-FROM,
3776 thanks to some patches originally by Tim Moore.
3777 ** The system no longer hangs when dumping circular lists to fasl
3778 files, thanks to a patch originally from Douglas Crosher.
3779 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3780 undefined function error.
3781 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3782 in structure slot initforms) for now, documented workaround instead:-|
3783 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3784 * fixed bug 5: added stubs for various Gray stream functions called
3785 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3786 installed, at least appropriate type errors are generated
3787 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3788 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3789 consistently in DEFMETHOD forms.
3790 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3791 it doesn't seem to affect SBCL after all
3792 * The C runtime system now builds with better optimization and many
3793 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3795 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3797 * The system is now under CVS at SourceForge (instead of the
3798 CVS repository on my home machine).
3799 * The new signal handling code has been tweaked to treat register
3800 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3801 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3802 downstream, e.g. in debug-int.lisp, has implicit dependencies
3803 on the unsignedness of integer representation of machine words,
3804 and that caused the system to bomb out with infinite regress
3805 when trying to recover from type errors involving signed values,
3806 e.g. (BUTLAST '(1 2 3) -1).)
3807 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3808 (This was one of the bugs Peter Van Eynde reported back in July.)
3809 * The system now uses code inspired by Colin Walters' O(N)
3810 implementation of MAP (from the cmucl-imp@cons.org mailing
3811 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3812 inline the MAP operation, and there is more than one
3813 sequence argument to the MAP call (so that it can't just
3814 do ETYPECASE once and for all based on the type of the
3815 single sequence argument). (The old non-inline implementation
3816 of the general M-argument sequence-of-length-N case required
3817 O(M*N*N) time when any of the sequence arguments were LISTs.)
3818 * The QUIT :UNIX-CODE keyword argument has been renamed to
3819 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3820 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3822 * Raymond Wiker's patches to port dynamic loading from Linux to
3823 FreeBSD have been added.
3824 * The BUGS file is now more nearly up to date, thanks in large part
3825 to Martin Atzmueller's review of it.
3826 * The debugger now flushes standard output streams before it begins
3827 its output ("debugger invoked" and so forth).
3828 * The core version number and fasl file version number have both
3829 been incremented, because of incompatible changes in the layout
3831 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3832 used to not be called for a saved Lisp image.)
3833 * Martin Atzmueller's version of a patch to fix a compiler crash,
3834 as posted on sbcl-devel 13 September 2000, has been installed.
3835 * Instead of installing Martin Atzmueller's patch for the
3836 compiler transform for SUBSEQ, I deleted the compiler transform,
3837 and transforms for some similar consing operations.
3838 * A bug in signal handling which kept TRACE from working on OpenBSD
3840 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3841 away by constant folding
3842 * The system now defines its address space constants in one place
3843 (in the Lisp sources), and propagates them automatically elsewhere
3844 (through GENESIS and the sbcl.h file). Therefore, patching the
3845 address map is less unnecessarily tedious and error-prone. The
3846 Lisp names of address space constants have also been systematized.
3847 * CVS tags like dollar-Header-dollar have been removed from
3848 the sources, because they have never saved me trouble and
3849 they've been source of trouble working with patches and other
3850 diff-related operations.
3851 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3852 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3854 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3856 * The system has been ported to OpenBSD.
3857 * The system now compiles with a simple "sh make.sh" on the systems
3858 that it's supported on. I.e., now you no longer need to tweak
3859 text in the target-features.lisp-expr and symlinks in src/runtime/
3860 by hand, the make.sh takes care of it for you.
3861 * The system is no longer so grossly inefficient when compiling code
3862 involving vectors implemented as general (not simple) vectors (VECTOR T),
3863 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3864 which dares to use the various sequence functions on non-simple
3865 vectors, takes less of a performance hit.
3866 * There is now a primitive type predicate VECTOR-T-P
3867 to test for the (VECTOR T) type, so that e.g.
3868 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3869 can now be compiled with some semblance of efficiency. (The old code
3870 turned the type declaration into a full call to %TYPEP at runtime!)
3871 * AREF on (VECTOR T) is still not fast, since it's still compiled
3872 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3873 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3874 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3875 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3876 without making full calls to SUBTYPEP at runtime.
3877 (Some analogous efficiency issues for non-simple vectors specialized to
3878 element types other than T, or for non-simple multidimensional arrays,
3879 have not been addressed. They could almost certainly be handled the
3880 same way if anyone is motivated to do so.)
3881 * The changes in array handling break binary compatibility, so
3882 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3883 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3884 instead of (VALUES T T).
3885 * By following the instructions that Dan Barlow posted to sbcl-devel
3886 on 2 July 2000, I was able to enable primitive dynamic object
3887 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3888 functionality is not implemented (since it calls ld to resolve
3889 library references automatically, requiring RUN-PROGRAM for its
3890 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3891 try to resolve library references) is now supported.
3892 * The system now flushes the standard output streams when it terminates,
3893 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3894 them at several other probably-convenient times, e.g. in each pass of
3895 the toplevel read-eval-print loop, and after evaluating a form given
3896 as an "--eval" command-line option. (These changes were motivated by a
3897 discussion of stream flushing issues on cmucl-imp in August 2000.)
3898 * The source transform for TYPEP of array types no longer assumes
3899 that an array whose element type is a not-yet-defined type
3900 is implemented as an array of T, but instead punts, so that the
3901 type will be interpreted at runtime.
3902 * There is now some support for cross-compiling in make.sh: each of
3903 the phases of make.sh has its own script. (This should be transparent
3904 to people doing ordinary, non-cross-compile builds.)
3905 * Since my laptop doesn't have hundreds of megabytes of memory like
3906 my desktop machine, I became more motivated to do some items on
3907 my to-do list in order to reduce the size of the system a little:
3908 ** Arrange for various needed-only-at-cold-init things to be
3909 uninterned after cold init. To support this, those things have
3910 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3911 symbols with such names are now uninterned after cold init).
3912 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3913 fasl files for cold load.
3914 ** Remove the old compiler structure pooling code (which used to
3915 be conditional on the target feature :SB-ALLOC) completely.
3916 ** Redo the representation of some data in cold init to be more compact.
3917 (I also looked into supporting byte compiled code at bootstrap time,
3918 which would probably reduce the size of the system a lot, but that
3919 looked too complicated, so I punted for now.)
3920 * The maximum signal nesting depth in the src/runtime/ support code has
3921 been reduced from 4096 to 256. (I don't know any reason for the very
3922 large old value. If the new smaller value turns out to break something,
3923 I'll probably just bump it back up.)
3924 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3926 * Many, many bugs reported by Peter Van Eynde have been added to
3927 the BUGS list; some have even been fixed.
3928 * While enabling dynamic object file loading, I tried to make the
3929 code easier to understand, renaming various functions and variables
3930 with less ambiguous names, and changing some function calling
3931 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3932 * While trying to figure out how to do the OpenBSD port, I tried to
3933 clean up some of the code in src/runtime/. In particular, I dropped
3934 support for non-POSIX signal handling, added various comments,
3935 tweaked the code to reduce the number of compilation warnings, and
3936 renamed some files to increase consistency.
3937 * To support the new automatic configuration functionality in make.sh,
3938 the source file target-features.lisp-expr has been replaced with the
3939 source file base-target-features.lisp-expr and the machine-generated
3940 file local-target-features.lisp-expr.
3941 * fixed a stupid quoting error in make.sh so that using CMU CL
3942 "lisp -batch" as cross-compilation host works again
3944 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3946 * DESCRIBE no longer tries to call itself recursively to describe
3947 bound/fbound values, so that it no longer fails on symbols which are
3948 bound to themselves (like keywords, T, and NIL).
3949 * DESCRIBE now works on generic functions.
3950 * The printer now prints less-screwed-up representations of closures
3951 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3952 * A private symbol is used instead of the :EMPTY keyword previously
3953 used to mark empty slots in hash tables. Thus
3954 (DEFVAR *HT* (MAKE-HASH-TABLE))
3955 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3956 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3957 now does what ANSI says that it should. (You can still get
3958 similar noncompliant behavior if bang on the hash table
3959 implementation with all the symbols you get back from
3960 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3961 This breaks binary compatibility, since tests for equality to
3962 :EMPTY are wired into things like the macroexpansion of
3963 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3965 * There's now a minimal placeholder implementation for CL:STEP,
3966 as required by ANSI.
3967 * An obscure bug in the interaction of the normal compiler, the byte
3968 compiler, inlining, and structure predicates has been patched
3969 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3970 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3972 * Missing ordinary arguments in a macro call are now detected even
3973 when the macro lambda list contains &KEY or &REST.
3974 * The debugger no longer complains about encountering the top of the
3975 stack when you type "FRAME 0" to explicitly instruct it to go to
3976 the top of the stack. And it now prints the frame you request even
3977 if it's the current frame (instead of saying "You are here.").
3978 * As specified by ANSI, the system now always prints keywords
3979 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3981 * The default initial SIZE of HASH-TABLEs is now smaller.
3982 * Type information from CLOS class dispatch is now propagated
3983 into DEFMETHOD bodies, so that e.g.
3984 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3986 is now basically equivalent to
3987 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3988 (DECLARE (TYPE SINGLE-FLOAT X))
3990 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3991 operation, without having to do run-time type dispatch.
3992 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3993 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3994 * Since I don't seem to be making big file reorganizations very often
3995 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3996 is overflowing my ability to conveniently back them up), I've finally
3997 checked the system into CVS. (The CVS repository is on my home system,
3998 not at SourceForge -- putting it on SourceForge might come later.)
3999 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4000 high-level GC-NOTIFY-FOO functions send their output. (There's
4001 still very little control of where low-level verbose GC functions
4002 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4003 less than it used to -- the GC-NOTIFY-FOO functions are now under
4004 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4005 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4006 in only one place in the source code, and propagates it automatically
4007 everywhere that it's needed. Thus e.g. when I bump the version from
4008 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4009 * The C source files now include boilerplate legalese and documentation
4010 at the head of each file (just as the Lisp source files already did).
4011 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4012 to his page will be replaced with a link to his new CLiki service.
4014 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4016 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4017 * The build process now looks for GNU make under the default name "gmake",
4018 instead of "make" as it used to. If GNU make is not available as "gmake"
4019 on your system, you can change this default behavior by setting the
4020 GNUMAKE environment variable.
4021 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4022 can build without error under CMU CL.
4024 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4026 * There is now a partial SBCL user manual (with some new text and some
4027 text cribbed from the CMU CL manual).
4028 * The beginnings of a profiler have been added (starting with the
4029 CMU CL profiler and simplifying and cleaning up). Eventually the
4030 main interface should be through the TRACE macro, but for now,
4031 it's still accessed through vaguely CMU-CL-style functions and macros
4032 exported from the package SB-PROFILE.
4033 * Some problems left over from porting CMU CL to the new
4034 cross-compilation bootstrap process have been cleaned up:
4035 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4036 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4037 part when getting rid of the compiler *BACKEND* stuff.)
4038 ** The value of *NULL-TYPE* was screwed up, because it was
4039 being initialized before the type system knew the final
4040 definition of the 'NULL type. This screwed up several key
4041 optimizations in the compiler, causing inefficiency in all sorts
4042 of places. (I found it because I wanted to understand why
4043 GET-INTERNAL-RUN-TIME was consing.)
4044 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4045 PROCLAIM FTYPE information. Unfortunately this broke binary
4046 compatibility again, since now the forms output by DEFGENERIC
4047 to refer to functions which didn't exist in 0.6.3.
4048 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4049 can use the new (as of 0.6.3) transform for SOME into MAP into
4051 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4052 values to (INTEGER 0 1000000), so that the time code will no longer
4053 occasionally get blown up by Linux returning 1000000 microseconds
4054 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4055 its output conform to the ANSI spec. (Alas, this makes its output
4056 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4057 * A full call to MAP NIL with a single sequence argument no longer conses.
4058 * fixes to problems pointed out by Martin Atzmueller:
4059 * The manual page no longer talks about multiprocessing as though
4060 it were currently supported.
4061 * The ILISP support patches have been removed from the distribution,
4062 because as of version 5.10.1, ILISP now supports SBCL without us
4063 having to maintain patches.
4064 * added a modified version of Raymond Toy's recent CMU CL patch for
4065 EQUALP comparison of HASH-TABLE
4067 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4069 * The system still can't cross-compile itself with
4070 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4071 entails), but at least it can compile more of itself that way
4072 than it used to be able to, and various buglets which were uncovered
4073 by trying to cross-compile itself that way have now been fixed.
4074 * This release breaks binary compatibility again. This time
4075 at least I've incremented the FASL file format version to 2, so that the
4076 problem can be detected reliably instead of just causing weird errors.
4077 * various new style warnings:
4078 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4079 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4080 ** using the lexical binding of a variable named in the *FOO* style
4081 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4082 as specified by ANSI.
4083 * *RANDOM-STATE* is no longer automatically initialized from
4084 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4085 default behavior of the system is to repeat its behavior every time
4086 it's run. If you'd like to change this behavior, you can always
4087 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4088 old convention there was no comparably easy way to get the system to
4089 repeat its behavior every time it was run.
4090 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4091 FTYPE declarations has been removed, in favor of their ANSI
4092 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4093 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4094 the types of their sequence arguments can be determined at compile time.
4095 This is done through a new open code expansion for MAP which eliminates
4096 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4097 argument types can be determined at compile time.
4098 * The optimizer now transforms COERCE into an identity operation when it
4099 can prove that the coerced object is already of the correct type. (This
4100 can be a win for machine generated code, including the output of other
4101 optimization transforms, such as the MAP transform above.)
4102 * Credit information has been moved from source file headers into CREDITS.
4103 * Source file headers have been made more standard.
4104 * The CASE macro now compiles without complaining even when it has
4107 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4109 * (Note that the way that the PCL macroexpansions were rewritten
4110 to accommodate the change in DEFGENERIC below breaks binary
4111 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4112 not run under sbcl-0.6.2. Once we get out of alpha releases,
4113 i.e. hit release 1.0.0, we'll probably try to maintain binary
4114 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4115 and sbcl-1.4.4. Until then, however, it might be fairly common
4116 for maintenance releases to break binary compatibility.)
4117 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4119 * The reporting of unhandled signals has been changed to print some
4120 explanatory text as well as the report form. (Previously only
4121 the report form was printed.)
4122 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4123 it defines, so that the compiler no longer issues undefined function
4124 warnings for compiled-but-not-yet-loaded generic functions.
4125 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4126 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4127 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4128 section of the ANSI spec are supported.
4129 * The "" nickname for the "KEYWORD" package has been removed.
4130 The reader still handles symbol tokens which begin with a package marker
4131 as keywords, but it doesn't expose its mechanism for doing so in the
4132 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4133 * The system now issues STYLE-WARNINGs for contradictory TYPE
4134 proclamations. (Warnings for contradictory FTYPE proclamations would
4135 be nice too, but those can't be done usefully unless the type system
4136 is made smarter about FUNCTION types.)
4137 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4138 systematized, so that "*target-*.lisp is supposed to exist only on the
4139 target and imply that there's a related file which exists on the
4140 host, and *host-*.lisp is supposed to exist only on the host and imply
4141 that there's a related file which exists on the target. This involves a
4142 lot of renaming. Hopefully the acute confusion caused by the renaming
4143 will be justified by the reduction in chronic confusion..
4144 ** runtime-type.lisp -> early-target-type.lisp
4145 ** target-type.lisp -> late-target-type.lisp
4146 ** early-host-format.lisp -> early-format.lisp
4147 ** late-host-format.lisp -> late-format.lisp
4148 ** host-error.lisp -> misc-error.lisp
4149 ** early-error.lisp -> early-target-error.lisp
4150 ** late-error.lisp -> late-target-error.lisp
4151 ** host-defboot.lisp -> early-defboot.lisp
4152 ** code/misc.lisp -> code/target-misc.lisp
4153 ** code/host-misc.lisp -> code/misc.lisp
4154 ** code/numbers.lisp -> code/target-numbers.lisp
4155 ** code/early-numbers.lisp -> numbers.lisp
4156 ** early-host-type.lisp -> early-type.lisp
4157 ** late-host-type.lisp -> late-type.lisp
4158 ** host-typep.lisp -> typep.lisp
4159 ** load.lisp -> target-load.lisp
4160 ** host-load.lisp -> load.lisp
4161 ** host-disassem.lisp -> disassem.lisp
4162 ** host-insts.lisp -> insts.lisp
4163 ** byte-comp.lisp -> target-byte-comp.lisp
4164 ** host-byte-comp.lisp -> byte-comp.lisp
4165 ** host-signal.lisp -> signal.lisp
4166 ** host-defstruct.lisp -> defstruct.lisp
4167 ** late-target-type.lisp -> deftypes-for-target.lisp
4168 Furthermore, several other previously target-only files foo.lisp (e.g.
4169 hash-table.lisp and random.lisp) have been split into a target-and-host
4170 foo.lisp file and a target-only target-foo.lisp file, with their key type
4171 definitions in the target-and-host part, so that the cross-compiler will
4172 know more about target types.
4173 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4174 gone away. In their place are various *BACKEND-FOO* variables
4175 corresponding to the slots of the old structure.
4176 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4177 into the target SBCL has been fixed.
4178 * The chill.lisp system for loading cold code into a running SBCL
4180 * Support for the CMU CL "scavenger hook" extension has been removed.
4181 (It was undocumented and unused in the CMU CL sources that SBCL was
4182 derived from, and stale in sbcl-0.6.1.)
4183 * Various errors in the cross-compiler type system were detected
4184 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4185 (enabling various consistency checks). Many of them were fixed,
4186 but some hard problems remain, so the compiler is back to
4187 running without *TYPE-SYSTEM-INITIALIZED* for now.
4188 * As part of the cross-compiler type system cleanup, I implemented
4189 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4190 * I have started adding UNCROSS calls throughout the type system
4191 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4192 on cross-compiler input files will be able to go away, and various
4194 * CONSTANTP now returns true for quoted forms (as explicitly required
4197 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4199 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4200 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4201 (MAKE-LIST -1). In the longer term, it ought to become true
4202 that declarations are assertions even at SAFETY 1. For now, it's not
4203 quite true even at SAFETY 3, but it's at least more nearly true..
4204 (Note that this change seems to increases the size of the system by
4205 O(5%) and to decrease the speed of the compiler by 20% or more.)
4206 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4207 for the problem of printing dozens of lines of distracting information
4208 about low-level system machinery as part of the top stack frame
4209 on entry to the debugger when an undefined function was called.
4210 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4211 is not reset to COMMON-LISP-USER.
4212 * Compilation of stuff related to dyncount.lisp has been made conditional
4213 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4214 smaller. The various dyncount-related symbols have been moved into
4215 a new "SB-DYNCOUNT" package.
4216 * tty-inspect.lisp has been renamed to inspect.lisp.
4217 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4218 feature has gone away. (When we eventually port to other flavors of
4219 libc and/or Unix, we'll try to make the differences between flavors
4220 invisible at the user level.)
4221 * Various other *FEATURES* tags, and/or their associated conditionals,
4222 have been removed if obsolescent, or given better documentation, or
4223 sometimes given more-mnemonic names.
4225 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4227 * tidied up "make.sh" script
4228 * tidied up system directory structure
4229 * better "clean.sh" behavior
4230 * added doc/FOR-CMUCL-DEVELOPERS
4231 * many many small tweaks to output format, e.g. removing possibly-confusing
4232 trailing #\. character in DESCRIBE-INSTANCE
4233 * (EQUALP #\A 'A) no longer signals an error.
4234 * new hashing code, including EQUALP hashing
4235 * tidied up Lisp initialization and toplevel
4236 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4237 * command line argument processing
4238 * added POSIX-GETENV function to deal with Unix-ish environment variables
4239 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4240 terminating SBCL on EOF
4241 * non-verbose GC by default
4242 * There is no more "sbcl" shell script; the sbcl file is now the C
4243 runtime executable (just like CMU CL).
4244 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4246 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4247 symbol and package information as Lisp native symbol and package objects
4248 instead of strings naming symbols and strings naming packages. This way,
4249 whenever packages are renamed (as in warm init), debug information is
4250 transformed along with everything else.
4251 * tweaked the optimization policy declarations which control the building
4252 of SBCL itself. Now, among other things, the system no longer saves
4253 source location debugging information. (This helps two problems at once
4254 by reducing SBCL size and by keeping SBCL from trying to look for its
4255 sources -- which may not exist -- when reporting errors.)
4256 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4257 debugging and testing purposes
4258 * cleaned up printing, making the printer call PRINT-OBJECT for
4259 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4260 methods, giving nearly-ANSI behavior
4261 * converted almost all special variables to use *FOO* naming convention
4262 * deleted PARSE-TIME functionality, since it can be done portably
4263 * moved some files out of cold init into warm init
4264 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4266 * regularized formatting of source files
4267 * added an install.sh script
4268 * fixed ridiculous memory usage of cross-compiler by making
4269 compiler/alloc.lisp not try to do pooling unless it can hook
4270 itself into the GC of the cross-compilation host. Now the system
4271 builds nicely on my old laptop.
4272 * added :SB-ALLOC in target-features.lisp-expr
4273 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4274 was not implemented)
4275 * re-did condition handling and note reporting in the compiler. Notes
4276 are no longer handled by signalling conditions. Style warnings
4277 and warnings are handled more correctly and reported in such a way
4278 that it's easy to find one or the other in your output (so that you
4279 can e.g. figure out which of many problems caused COMPILE-FILE to
4281 * changed the severity of several compiler warnings from full WARNING
4282 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4283 compiler note reporting so that it doesn't use the condition system
4284 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4285 COMPILE-FILE command)
4286 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4287 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4288 no longer supported, since their implementation was deeply intertwingled
4289 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4290 * removed bogus "support" for compiler macros named (SETF FOO), and
4291 removed the compiler macro for SETF INFO (but only after making a fool
4292 of myself on the cmucl-imp mailing list by posting a bogus patch for
4293 DEFINE-COMPILER-MACRO..)
4294 * Compiled files containing forms which have side effects on the Lisp
4295 reader (such as DEFPACKAGE forms) are now handled more correctly.
4296 (Compiler queuing of top level lambdas has been suppressed by setting
4297 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4298 * deleted various currently-unused source files, e.g. gengc.lisp. They
4299 may be added back at some point e.g. when porting to other architectures,
4300 but until they are it's distracting to distribute them and to try to
4302 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4303 there were so many of them they're just distractions, and UNCROSS is
4304 known to be able to handle the current sources
4305 * moved PROFILE functionality into TRACE, so that it will be clear
4306 how the wrapping and unwrapping of functions when you profile them
4307 interacts with the wrapping and unwrapping of functions when you
4308 trace them. (Actually, the functionality isn't there yet, but at least
4309 the interface specification is there. Hopefully, the functionality will
4310 arrive with some future maintenance release.)
4311 * removed host-oops.lisp
4312 * changed signature of QUIT function to allow UNIX-CODE argument
4313 * fixed READ-SEQUENCE bug
4314 * tweaked verbose GC output so that it looks more like the progress
4315 output that ANSI specifies for functions like LOAD
4316 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4317 * added <http://sbcl.sourceforge.com> to the banner information printed by