1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
3 * new contrib: sb-cover, an experimental code coverage tool, is included
5 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
7 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
8 produce more efficient code on x86-64 if the compiler can determine
9 that (+ INDEX <constant>) does not require a bounds check and FOO
10 has an element type at least 8 bits wide.
11 * enhancement: a new, experimental synchronous timeout facility is
12 provided. Refer to SB-SYS:WITH-DEADLINE for details.
13 * enhancement: when a symbol name conflict error arises, the
14 conflicting symbols are always printed with a package prefix.
15 (thanks to Kevin Reid)
16 * enhancement: stepping is now once again supported on the SPARC and
17 MIPS platforms. (It is also now more likely to work on CheneyGC
19 * enhancement: sb-sprof can now also track and report accurate call
21 * bug fixes: the treatment of non-standard subclasses of
22 SB-MOP:SPECIALIZER is more correct.
23 * incompatible change: PURIFY no longer copies the data from the
24 dynamic space into the static and read-only spaces on platforms
25 that use the generational garbage collector
26 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
28 * bug fix: GC race condition occasionally resulting in crashes with
29 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
31 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
33 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
34 a STYLE-WARNING to be signalled (regression from 1.0.4.)
35 * bug fix: an asynchronous interrupt could previously leave the
36 system running with GC inhibited.
37 * bug fix: a DECLARE form evaluated at top-level now causes an error
38 rather than silently (or verbosely) returning NIL.
39 * bug fix: trying to dissassemble functions compiled at high DEBUG could
40 result in a error being signalled due to source form lookup errors.
41 (reported by Peter Graves)
43 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
44 * incompatible change: removed writer methods for host-ent-name,
45 host-ent-addresses -- changing the values did not update the DNS
47 * minor incompatible change: changed experimental JOIN-THREAD interface
48 * documentation: the manual now lists reader and writer methods
49 in class slot documentation sections. (thanks to Richard M Kreuter)
50 * documentation: unwinding from asyncronous events has been
52 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
53 in multithreaded application code.
54 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
55 platforms. (thanks to James Anderson for the optimization hint)
56 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
58 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
59 variants no longer cons.
60 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
61 their NOT- variants no longer cons.
62 * optimization: EQUAL hash tables no longer use SXHASH for objects
63 of all data types, but instead use an EQL hash for types for which
64 EQUAL is the same as EQL.
65 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
66 are significantly faster.
67 * optimization: new STRING-OUTPUT-STREAM implementation is much
68 faster for large outputs and conses 30% less on average.
69 * enhancement: XREF information is now collected for references made
70 to global variables using SYMBOL-VALUE and a constant argument.
71 * enhancement: SIGINT now causes a specific condition
72 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
73 * bug fix: FILE-POSITION used to signal an error in some cases where
74 ANSI requires it to return NIL.
75 * bug fix: ADJUST-ARRAY is now interrupt-safe.
76 * bug fix: adding and removing fd-handlers is now interrupt-safe.
77 * bug fix: inlined calls to C now ensure 16byte stack alignment on
79 * bug fix: bad type declaration in the CLOS implementation has
80 been fixed. (reported by James Anderson)
81 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
82 (thanks to Lutz Euler)
83 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
84 required by the CLHS. (reported by Samium Gromoff)
85 * bug fix: dead unbound variable references now signal an error.
86 (reported by Marco Monteiro)
87 * bug fix: / with an unused value was being deleted in safe code.
88 (thanks to Marco Monteiro and Kevin Reid)
89 * bug fix: number of characters that can be written onto a single
90 line in a file is unlimited.
91 * bug fix: some GC deadlocks caused by asynchronous interrupts have
92 been fixed by inhibiting interrupts for when GC is disbled.
93 * bug fix: some interrupt safety issues with GETHASH, (SETF
94 GETHASH), CLRHASH and REMHASH have been fixed.
95 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
96 specifier no longer causes infinite recursion.
97 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
98 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
99 * bug fix: an &environment argument with the correct variable information
100 is passed to macros that are expanded during byte compilation.
101 (reported by Samium Gromoff)
102 * bug fix: a logic error was causing the hash function for bignums to
103 have excessive amounts of collisions. (reported by Faré Rideau)
104 * bug fix: modifying the contents of an array could change the return
105 value of SXHASH on that array, which is only allowed for strings
106 and bit vectors. (bug introduced in 0.9.16)
107 * improvement: the x86-64/darwin port now passes all tests (except
108 for the debugger tests) but should still be considered
109 experimental until this is fixed.
110 * improvement: a style-warning is signaled for CASE (etc) clauses with
111 duplicate keys. (thanks to Kevin Reid)
112 * improvement: macos/x86 and macos/x86-64 now use mach exception
113 handlers for dealing with illegal instructions (for trapping and
114 error handling) and memory protection violations (for GC).
116 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
117 * new platform: experimental support for x86-64/darwin (MacOS).
118 * incompatible change: the thread-safe (on most platforms) getaddrinfo
119 and getnameinfo sockets functions are used instead of gethostbyaddr
120 and gethostbyname, on platforms where the newer functions are available.
121 As a result, the ALIASES field of HOST-ENT will always be NIL on these
123 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
124 don't print the startup banner, but behave as if --noinform was passed
125 as a command line argument. (thanks to Kevin Reid)
126 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
127 * optimization: code using alien values with undeclared types is much faster.
128 * optimization: the compiler is now able to open code SEARCH in more cases.
129 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
130 * bug fix: using standardized COMMON-LISP special variables as loop
131 variables no longer signals bogus package lock violations. (reported
133 * bug fix: declaring local loop variables to be of a range-limited type
134 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
135 (reported by Andras Simon)
136 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
137 bugs remain on x86-64.)
138 * bug fix: the #= and ## reader macros now interact reasonably with
139 funcallable instances.
140 * bug fix: type-checks for function arguments were compiled using the
141 compiler policy settings of the wrong lexical environment (in 1.0.2
143 * bug fix: SHADOW accepts characters as string designators, as required
144 by the spec (thanks to Eric Marsden)
145 * bug fix: fixed GC safety issues when foreign functions are called with
146 non-base strings as arguments
147 * bug fix: more consistent error output for fatal-compiler-warnings, like
149 * bug fix: functions with proper names were showing up as NIL in
150 backtraces (reported by Edi Weitz, regression in 1.0.2)
152 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
153 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
154 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
155 produce more efficient code on the x86 if the compiler can determine
156 that (+ INDEX <constant>) does not require a bounds check and FOO
157 has an element type at least 8 bits wide.
158 * bug fix: references to undefined variables are handled the same way
159 in toplevel forms as in the normal compiler
160 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
162 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
163 (thanks to Jon Buller)
164 * improvement: faster compilation times for complex functions
165 * improvement: added readlink support to SB-POSIX (thanks to Richard
168 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
169 * improvement: experimental support for mach exception handling on
170 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
172 * improvement: support for GBK external format.
173 (thanks to Chun Tian (binghe))
174 * improvement: the debugger now displays variables that have been closed
175 over, in code compiled with (DEBUG 2) or higher
176 * improvement: support for executable cores on NetBSD (thanks to
178 * new feature: added a RESTART-FRAME debugger command
179 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
180 be used to provide an implementation for FILE-POSITION on Gray streams
181 (thanks to Eric Marsden)
182 * new feature: add syslog, openlog and closelog support to SB-POSIX
183 (thanks to Richard Kreuter)
184 * optimization: the function call overhead in code compiled with
185 a high DEBUG optimization setting is significantly
186 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
187 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
188 (thanks to Tony Martinez)
189 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
190 evaluated once (reported by Kevin Reid)
191 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
192 works again (thanks to Kilian Sprotte)
193 * bug fix: an error is signaled for tagbodies with duplicate tags
194 (thanks to Stephen Wilson)
195 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
196 * bug fix: Win32 port can now handle foreign code unwinding Lisp
197 stack frames from alien callbacks.
198 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
199 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
200 65k pages (thanks to David Woodhouse)
201 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
203 changes in sbcl-1.0.1 relative to sbcl-1.0:
204 * new platform: FreeBSD/x86-64, including support for threading.
205 * new feature: the compiler stores cross-referencing information
206 abount function calls (who-calls), macroexpansion (who-macroexpands)
207 and special variables (who-binds, who-sets, who-references) for code
208 compiled with (< SPACE 3). This information is available through the
209 sb-introspect contrib.
210 * new feature: users may subclass SEQUENCE, and have instances of
211 these classes interoperate with standard Common Lisp functions if
212 a number of methods are defined. (This feature is experimental
213 and the interface subject to change based on feedback from SBCL
214 users and the general community)
215 * improvement: sb-sprof traces call stacks to an arbitrary depth on
216 x86/x86-64, rather than the previous fixed depth of 8
217 * improvement: another pthread back-end of mutex "pthread-futex".
218 * bug fix: non-ascii command-line arguments are processed correctly
219 (thanks to Yaroslav Kavenchuk)
220 * bug fix: non-required arguments were not passed correctly when a method
221 defined using DEFMETHOD was called from a mop-generated method using
222 CALL-NEXT-METHOD (reported by Pascal Costanza)
223 * bug fix: recursion is now permitted in accessors through
224 SLOT-UNBOUND. (reported by Pascal Costanza)
225 * bug fix: an error was signaled at startup if the HOME environment
226 variable was defined, but had an empty value (reported by Peter Van Eynde)
227 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
228 signal a PROGRAM-ERROR, not a BUG.
229 * bug fix: SB-POSIX:READDIR works when built with large file support.
230 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
232 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
233 for better automatic indentation support. (thanks to Matt Pillsbury)
234 * optimization: loading generic functions no longer takes O(n^2) time,
235 proportional to the amount of methods in the generic function
236 (reported by Todd Sabin and Jeremy Brown)
237 * optimization: the FIND and POSITION family of sequence functions
238 are significantly faster on arrays whose element types have been
240 * improvements to the Windows port:
241 ** Intermittent heap corruption problems have been fixed. (thanks
242 to Alastair Bridgewater)
243 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
245 ** Lisp is able to unwind foreign exception frames from alien
246 callbacks. (thanks to Alastair Bridgewater)
248 changes in sbcl-1.0 relative to sbcl-0.9.18:
249 * improvement: experimental support for threading on FreeBSD/x86.
250 (thanks to NIIMI Satoshi)
251 * improvement: runtime option --dynamic-space-size can be used
252 to set the size of the dynamic space reserved on startup.
253 * improvement: floating point modes in effect are now saved in
254 core, and restored on startup.
255 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
256 startup, not time since first call to GET-INTERNAL-REAL-TIME.
257 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
258 threads are not running after *SAVE-HOOKS* have run.
259 * improvement: writes to CLOS instance slots are type-checked in code
260 compiled with (SAFETY 3)
261 * improvement: floating-point exception handling on FreeBSD (thanks to
263 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
264 (thanks to Zach Beane)
265 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
267 * improvement: added support for the Shift-JIS external format.
268 (contributed by NIIMI Satoshi)
269 * improvement: callbacks are supported on Linux/PPC. (thanks to
271 * bug fix: compiler bug triggered by a (non-standard) VALUES
272 declaration in a LET* was fixed. (reported by Kaersten Poeck)
273 * bug fix: file compiler no longer confuses validated and already
274 dumped structurres. (reported by Kaersten Poeck)
275 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
276 fill-pointer signals a type-error as required. (thanks to
278 * bug fix: disassemly of funcallable instances works.
279 * bug fix: single stepping on PPC.
280 * bug fix: fix thread-safety problems in the type system (generally
281 manifesting as nonsensical errors like "STRING is a bad type specifier
282 for sequences" or "The value 1 is not of type FIXNUM")
283 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
284 (reported by Josip Gracin)
285 * bug fix: an error is signaled for attempts to displace arrays with
286 incompatible element types (thanks to Mario Mommer)
287 * bug fix: more correct handling of wide characters in debug info
288 (bug reported by Attila Lendvai and fixed by Juho Snellman)
289 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
290 and don't cause extra consing
291 * optimization: MAP and MAP-INTO are significantly faster on vectors
292 whose elements types have been declared.
293 * Improvements to SB-SPROF:
294 ** Support for allocation profiling
295 ** Reduced profiling overhead, especially for long profiling runs
296 * Improvements to the Windows port:
297 ** floating point exceptions are now reported correctly.
298 ** stack exhaustion detection works partially.
299 ** more accurate GET-INTERNAL-REAL-TIME.
300 ** better RUN-PROGRAM behaviour in case of errors in the child
302 ** PROBE-FILE now simplifies pathnames correctly.
303 ** DIRECTORY now works correctly with :WILD-INFERIORS.
304 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
305 1st 1970 (although time timezone data will be unreliable).
306 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
308 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
309 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
310 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
312 * bug fix: two potential GC deadlocks affecting threaded builds.
313 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
314 returns T (reported by Anton Kazennikov)
315 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
316 with non-variable places
317 * bug fix: remove a race condition in the setting of
318 funcallable-instance functions, this should make threaded CLOS
319 code more stable against memory faults.
320 * bug fix: corruption of specials when unbinding is interrupted by an
321 asynchronous unwind (reported by Hannu Koivisto)
322 * improvement: the debugger will now also display local variables that
323 are only used once, for code compiled with a DEBUG optimization quality
326 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
327 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
328 * incompatible change: External-format support for FFI calls. The
329 SB-ALIEN:C-STRING no longer implies an ASCII
330 external-format. Instead the string is subject to external-format
331 conversion. Additionally return values of type SB-ALIEN:C-STRING
332 are by default of type (SIMPLE-ARRAY CHARACTER), not
333 SIMPLE-BASE-STRING. For an alien type that behaves like the old
334 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
335 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
336 * incompatible change: SB-EXT package no longer contains the
337 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
338 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
340 * incompatible change: the single-stepper is no longer available
341 on Alpha, Mips and Sparc platforms.
342 * minor incompatible change: the direct superclasses of
343 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
344 not (STANDARD-OBJECT FUNCTION). This makes the
345 class-precedence-lists of GENERIC-FUNCTION and
346 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
348 * minor incompatible change: the default stream external format on
349 non-unicode SBCL is no longer detected from the locale, but is always
351 * new feature: Add a version of evaluator that uses an interpreter instead
352 of the compiler. EVAL still uses the compiler by default, to switch it
353 to use the interpreter, set the value of the variable
354 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
355 * minor incompatible change: the single-stepper REPL has been merged
356 with the normal debugger (see the "Stepping" heading of the debugger help
357 for more details). The debugger command STEP will no longer switch
358 to the single-stepper REPL.
359 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
360 for a type now works.
361 * bug fix: loading of default sysinit file works. (thanks to Leonid
363 * bug fix: better detection of circularities in the file-compiler.
364 (reported by Marco Monteiro)
365 * bug fix: the CL pathname functions now work with files that have
366 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
367 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
368 systems with Unix98 pty semantics.
369 * bug fix: ASDF-INSTALL will now work with bsd tar.
370 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
372 * bug fix: timers expiring in dead threads no longer cause a
373 type-error (reported by Paul "Nonny Mouse").
374 * bug fix: thanks to more lightweight single-stepper instrumentation,
375 code compiled with (DEBUG 3) will compile and execute significantly faster,
376 and will have more accurate type-inferencing than before
377 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
378 whose bindings are modified
379 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
380 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
381 ** replace dummy implementations of CL:MACHINE-INSTANCE and
382 CL:SOFTWARE-VERSION with proper version
384 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
385 * feature: implemented the READER-METHOD-CLASS and
386 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
387 as specified by AMOP.
388 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
390 * optimization: faster LOGCOUNT implementation on x86 and x86-64
391 (thanks to Lutz Euler)
392 * optimization: hashing of general arrays and vectors has been
393 improved. (reported by Any Fingerhut)
394 * enhancement: SB-INTROSPECT is now able to find definitions of
395 profiled functions. (thanks to Troels Henriksen)
396 * enhancement: compiler-macro expansion applies now to FUNCALL forms
397 as well, allowing compiler-macros for SETF-functions to expand.
398 * enhancement: step-instrumentation no longer wraps known
399 single-value functions with multiple-value context, allowing
400 better type inference.
401 * fixed bug #337: use of MAKE-METHOD in method combination now works
402 even in the presence of user-defined method classes. (reported by
403 Bruno Haible and Pascal Costanza)
404 * fixed bug #339(c): if there are applicable methods not part of any
405 long-form method-combination group, call INVALID-METHOD-ERROR.
406 (reported by Bruno Haible)
407 * fixed bug #361: the :FUNCTION initarg in the protocol for
408 initialization of methods can now be used to override
409 internally-produced optimized functions. (reported by Bruno
411 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
412 system-provided lambda expression no longer cause warnings about
413 unbound #:|pv-table| symbols.
414 * bug fix: improved the handling of type declarations and the
415 detection of violations for keyword arguments with non-constant
417 * bug fix: potentially erronous calls to PATHNAME and
418 MERGE-PATHNAMES were being flushed in some cases.
419 (reported by Richard Kreuter)
420 * bug fix: compiled calls to TYPEP were mishandling obsolete
421 instances. (reported by James Bielman and Attila Lendvai)
422 * bug fix: format strings with ~> without matching ~< no longer
423 trigger an AVER, but signal an understandable error instead.
424 (reported by Antonio Martinez)
425 * bug fix: specifying an output-file without a pathname-type for
426 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
427 of input-file instead of "fasl". (reported by Robert Dodier)
428 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
429 (reported by James Y Knight).
430 * bug fix: compiler-macros-function did not consider the environment
431 argument for shadowing by local functions.
432 * bug fix: compiler-macros expansion was inhibited by local INLINE
434 * bug fix: inline expansions of known functions were subject to
435 step-instrumentation in high debug policies, leading to problems
437 * bug fix: compiler failed to differentiate between different CONS
439 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
440 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
441 * bug fix: #368: incorrect use of expressed vs. upgraded array
443 * bug fix: #306a: more precise unions of array types.
444 * thread-safety improvements:
445 ** CONDITION-WAIT could return early on Linux, if the thread was
446 interrupted and subsequently continued with SIGCONT.
447 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
449 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
450 * added support for the ucs-2 external format. (contributed by Ivan
452 * minor incompatible change: pretty printing of objects of type
453 (cons symbol) is, in the default pprint-dispatch-table, now
454 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
456 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
457 called later in a class's lifetime, possibly as late as when the
458 first instance of the class is created. Previously,
459 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
460 class became finalizeable.
461 * fixed bug: FILE-POSITION sometimes returned inconsistent results
462 for multibyte external-format streams. (thanks to "vbzoli")
463 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
464 with :ALLOCATION :CLASS inherited from superclasses of the
466 * fixed bug: anonymous classes can now be created using the :NAME
467 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
468 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
469 * fixed bug: core-files saved with :EXECUTABLE T can again be
470 executed when SBCL_HOME isn't set. (reported by James Knight)
471 * fixed bug: toplevel LOCALLY forms with declarations could
472 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
473 * fixed bug: printing from several different threads using different
474 values of *print-case* could cause invalid output, due to
475 some internal special variables of the printer not being bound
476 thread-locally (reported by Max Mikhanosha)
477 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
478 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
479 (reported by Pascal Costanza)
480 * fixed bug: FINALIZE-INHERITANCE is called from
481 REINITIALIZE-INSTANCE on classes when the class has previously
482 been finalized, as required by AMOP.
483 * minor code generation optimizations:
484 ** better register allocation in CLOS dispatching functions
485 ** overflow detection when coercing signed bytes to fixnums on x86-64
486 ** is now implemented with one IMUL instruction instead of three shifts
487 ** more efficient bit-vector access on x86 and x86-64
488 ** more efficient access to raw structure slots on x86 and x86-64
489 * fixed some bugs revealed by Paul Dietz' test suite:
490 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
493 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
494 * feature: thread support on Solaris/x86, and experimental thread support
496 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
498 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
499 default method for SLOT-UNBOUND.
500 * minor incompatible change: prevent the user from specializing the
501 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
502 somewhat counter to the intent of the protocol, I (CSR) think, and
503 additionally it just doesn't work in SBCL as currently
504 implemented, thanks to optimizations (that are always valid for
505 the other three SLOT-VALUEish functions, but not for the setter).
506 * minor incompatible change: the :SB-LDB feature is now enabled by
507 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
508 the low-level debugger.
509 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
510 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
511 down to an SB-SYS:FD-STREAM, allowing direct communication in
512 more cases. Temporary files and pipes are still used as a fallback
514 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
515 on the non-reentrant C functions localtime(3) and gmtime(3).
516 * bug fix: global optimization policy was not visible in LOCALLY and
518 * bug fix: class objects can be used as specializers in methods.
519 (reported by Pascal Costanza)
520 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
521 with a type error. (reported by blitz_ on #lisp)
522 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
523 failed with a type error. (reported by Pascal Bourguignon)
524 * bug fix: merging pathnames against defaults with :DIRECTORY
525 starting with '(:RELATIVE :BACK) should preserve the :BACK.
526 (reported by James Y Knight)
527 * bug fix: saving large (>2GB) cores on x86-64 now works
528 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
529 constant index and a (simple-array (signed-byte 32)) array
530 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
531 error (patch by Robert J. Macomber)
532 * feature: TIME now displays the amount of run-time spent in GC
533 * bug fix: The debugger now does a better job of respecting
534 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
535 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
536 workaround for bug 403.)
537 * bug fix: floating point exception handling now works on FreeBSD
538 (thanks to NIIMI Satoshi)
539 * fixed some bugs revealed by Paul Dietz' test suite:
540 ** MISC.641: LET-conversion were not supposed to work in late
542 * improvements to the Win32/x86 port:
543 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
544 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
546 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
547 * new feature: source path information is generated for macro-expansion
548 errors for use in IDE's like Slime (thanks to Helmut Eller)
549 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
550 * bug fix: compiler does not loop forever on an invalid type in
552 * improvement: compilation of most CLOS applications is significantly
554 * optimization: added a limited bytecode compiler for simple toplevel
555 forms, speeding up compilation and FASL loading
556 * bug fix: the statistical profiler now properly distinguishes anonymous
559 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
560 * minor incompatible change: in sbcl-0.9.11 (but not earlier
561 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
562 into subclasses of the finalized class. Now user calls to
563 FINALIZE-INHERITANCE finalize just the one class, and calls by the
564 system return before any subclasses are finalized.
565 * minor incompatible change: The reader no longer ignores errors
566 regarding non-existent packages in #+ and #- feature tests.
567 * new feature: command line options --no-sysinit, --no-userinit to
568 inhibit loading the corresponding init files
569 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
570 for sending data through UDP sockets (thanks to François-René Rideau)
571 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
572 error is signalled instead (thanks to François-René Rideau)
573 * minor incompatible change: Error signalling behaviour of lexical
574 operations violating package locks has changed slightly. Refer to
575 documentation on package locks for details.
576 * enhancement: EVAL can process IF-expressions without resorting to the
578 * bug fix: Some operations involving SIN, COS, or TAN weren't being
579 constant-folded properly due to missing out-of-line functions.
580 (reported by Mika Pihlajamäki)
581 * bug fix: LISTEN sometimes returned T even in cases where no data was
582 immediately available from the stream
583 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
584 were reversed. (reported by Levente Mészáros)
585 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
586 when given a vector displaced to an adjustable array. (reported
588 * bug fix: derivation of float boundaries from numbers outside the
589 appropriate float range (reported by John Wiseman)
590 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
592 * bug fix: precomputing cache entries for generic functions with
593 some subclasses of specializers as yet invalid does not attempt to
594 fill a cache line with a negative offset. (reported by Levente
596 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
597 allow condition class objects as arguments to DOCUMENTATION and
598 (SETF DOCUMENTATION); only find and set documentation for
599 structure names for the STRUCTURE doc-type. (suggested by Gary
601 * improvements to the Win32/x86 port:
602 ** (user-homedir-pathname) and default initialization file
603 locations now know about the user's "Documents and Settings"
604 directory (thanks to Yaroslav Kavenchuk)
605 ** run-program is implemented (thanks to Mike Thomas)
606 ** sockets support (thanks to Timothy Ritchey)
607 ** better backtrace support (thanks to Alastair Bridgewater)
608 ** sb-grovel supported
609 ** asdf-install and sb-posix work somewhat
610 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
611 * improvements to the Solaris/x86 port:
612 ** works on Solaris 11/Solaris Express
613 ** floating-point exception handling support
614 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
615 * fixed some bugs revealed by Paul Dietz' test suite:
616 ** REMOVE-METHOD returns its generic function argument even when
617 the method is not one of the generic functions' methods.
618 ** objects declared as MEMBER types can be admissible arguments to
620 ** printing characters should simply be printed by the FORMAT ~:C
622 ** compiler failure when compiling functions with hairy constant
623 defaults for optional parameters.
624 ** compiler produces wrong code when MAYBE-INLINE-expanding a
625 function, which is already optimized.
627 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
628 * new platform: experimental support for SBCL x86/Darwin, including
629 MacOS X 10.4.5 on Intel.
630 * new feature: Unicode character names are now known to the system
631 (through CHAR-NAME and NAME-CHAR).
632 * new feature: the filesystem location of SBCL's core file is
633 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
634 * minor incompatible change: the contrib modules SB-POSIX and
635 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
636 this change is to make it easier to distribute
637 location-independent binaries.
638 * bug fix: as implied by AMOP, standardized classes no longer have
639 slots named by external symbols of public packages. (reported by
641 * optimization: calling functions via constant symbols -- as in
642 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
643 via the function object as in (FUNCALL #'FOO).
644 * optimization: on x86 and x86-64, the calling convention for the
645 case of dealing with an unknown number of values has been altered
646 to be friendlier to the prediction heuristics implemented,
647 particularly with reference to CALL and RET pairing. (thanks to
648 Alastair Bridgewater)
649 * optimization: on x86, the code for access to thread-local symbol
650 values and binding/unbinding of thread-local symbols is smaller.
651 (thanks to Alastair Bridgewater)
652 * enhancement: CONSTANTP is now able to determine constantness of
653 more complex forms, including calls to constant-foldable standardized
654 functions and some special forms beyond QUOTE.
655 * fixed bug: occasional GC crashes on Solaris/x86
656 * optimization: x86-64 supports stack allocation of results of simple
657 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
658 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
659 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
660 (thanks to James Knight)
661 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
662 by the standard (thanks to Stephen Compall)
664 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
665 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
666 be used for bundling the runtime and the core file into one
667 executable binary. This feature is not currently supported on all SBCL
668 platforms. (thanks to James Bielman and NIIMI Satoshi)
669 * new feature: a generational or ephemeral garbage collector is now
670 the default on the PowerPC platform (both Linux and Darwin). The
671 old Cheney (stop and copy) collector is a build-time option.
672 (thanks to Cyrus Harmon, after Raymond Toy)
673 * minor incompatible change: the method by which SBCL finds its
674 contributed modules has changed; it no longer relies on symbolic
675 links from an $SBCL_HOME/systems directory, but searches directly
676 in the subdirectories of $SBCL_HOME.
677 * enhancement: the dynamic heap size on the Linux/PPC platform is
678 markedly larger, even using the older Cheney garbage collector.
679 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
680 certain complicated string types. (reported by Gary King)
681 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
682 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
683 * fixed bug: DOCUMENTATION on structures no longer signals an
684 error if no structure type of the right name exists. (reported by
686 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
687 applicable non-standard primary or :AROUND method on
688 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
689 no longer cause unbound variable errors. (reported by Kevin Reid)
690 * fixed bug: metacircle resolution in cases where methods have slots
691 added before the slots from STANDARD-METHOD. (reported by Jean
693 * fixed bug: the Power PC assembly code for calling into lisp
694 sometimes computed its return address wrongly (depending on the
695 15th bit of the address, fixed at link time). This used to
696 manifest itself in a segmentation violation while building PCL.
697 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
698 many others over the years)
699 * fixed bug: ROOM no longer reports silly numbers for stack usage if
700 the stack pointer (treated as a signed integer) is negative.
701 (thanks to Peter van Eynde)
703 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
704 * new platform: experimental support for the Windows operating
705 system has been added. (thanks to Alastair Bridgewater)
706 * fixed several bugs in and robustified the PPC FFI (including
707 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
708 * bug fix: allow non-simple string symbol names (reported by Paul
710 * bug fix: interrupt handling on NetBSD (thanks to Richard M
712 * bug fix: saving a core corrupted callbacks on x86/x86-64
713 * bug fix: closed a loophole in metacircularity detection and
714 grounding in the PCL implementation of CLOS.
715 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
716 (reported by tomppa on #lisp)
717 * bug fix: clear the direction flag on Lisp -> C transitions, as
718 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
719 (reported by Andrej Grozin and Hendrik Maryns)
720 * optimization: major improvements to GC efficiency on GENCGC platforms
721 * optimization: faster implementation of EQUAL
722 * optimization: emit more efficient opcodes for some common
723 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
724 * optimization: several other minor code-generation improvements on x86-64
725 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
727 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
728 * minor incompatible change: (SETF CLASS-NAME) and (SETF
729 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
730 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
731 by ANSI as a generic function, and for consistency (SETF
732 GENERIC-FUNCTION-NAME) is treated likewise.
733 * fixed bug #233.b: make constraint propagation notice when a variable
734 value is changed after it is referenced but before it is used
735 * fixed bug #296: no more arbitrary behaviour from filenames with
736 odd characters as --load arguments. (reported by Adam Warner)
737 * fixed bug #390: :CHARACTER-SET pathname components now work as
738 expected. (reported by Tim Daly Jr)
739 * fixed bug #391: complicated :TYPE intersections in slot
740 definitions no longer cause an error in PCL internals.
741 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
742 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
743 returning the number of octets which would be written to the
744 file-stream. (thanks to Robert J. Macomber)
745 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
746 arguments to the internal stream functions. (thanks to David
748 * bug fix: stack exhaustion is now handled on the x86/SunOS
749 platform. (thanks to Daisuke Homma)
750 * bug fix: more accurate ROOM results on GENCGC platforms
751 * optimization: improved type inference for arithmetic-for
752 index variables in LOOP
753 * optimization: faster floating-point SQRT on x86-64
754 * fixed some bugs revealed by Paul Dietz' test suite:
755 ** DOCUMENTATION returns NIL instead of "" for method combinations
756 that don't have a docstring
758 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
759 * minor incompatible change: (SETF CLASS-NAME) and (SETF
760 GENERIC-FUNCTION-NAME) are no longer generic functions, and
761 therefore are not customizeable by user code (as seems to be at
762 least permitted and maybe required by AMOP). As a consolation,
763 however, the SBCL implementation of these functions now calls
764 REINITIALIZE-INSTANCE as specified by AMOP.
765 * bug fix: slot-definition documentation is propagated as per ANSI
766 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
767 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
768 argument, as do eslotd initialization methods. (from Pascal
769 Costanza's "Closer" project)
770 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
771 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
773 * bug fix: it is now possible to have more than one subclass of
774 STANDARD-GENERIC-FUNCTION without causing stack overflow.
775 (reported by Bruno Haible, Pascal Costanza and others)
776 * bug fix: the dependent update protocol now works for generic
777 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
779 * bug fix: condition-class instances corresponding to
780 DEFINE-CONDITION forms are now created eagerly. (reported by
781 Kalle Olavi Niemitalo on comp.lang.lisp)
782 * bug fix: floating point printing is more accurate in some
783 circumstances. (thanks to Simon Alexander)
784 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
785 merged with *DEFAULT-PATHNAME-DEFAULTS*.
786 * bug fix: callbacks on OS X now preserve stack-alignment.
787 (thanks to Cyrus Harmon)
788 * enhancement: the x86-64 disassembler is much better at
789 disassembling SSE instructions. (thanks to Lutz Euler)
790 * enhancement: CHECK-TYPE on a variable now allows type-inference
791 (thanks to Vincent Arkesteijn)
792 * optimization: improved performance of EUC-JP external format.
793 (thanks to NIIMI Satoshi)
794 * optimization: performance improvements to IO on file streams of
795 :ELEMENT-TYPE CHARACTER
796 * optimization: much faster memory allocation on x86-64
797 * optimization: faster garbage collections (latency and throughput)
799 * optimization: faster bignum subtraction and division on x86 and x86-64
800 (thanks to Lutz Euler)
801 * optimization: more accurate type-inference for loops that use a
802 floating point index variable or a negative step.
804 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
805 * bug fix: add a workaround to SBCL looping infinitely at startup on
806 Linux kernels with apparently buggy implementations of personality().
807 (thanks to Svein Ove Aas)
808 * bug fix: Unicode symbols are correctly printed in LDB backtraces
809 (thanks to David Lichteblau)
810 * bug fix: local bindings shadow global symbol macros for
811 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
812 * optimization: non-open coded uses of numeric comparison operators
813 (e.g. >) no longer cons when called with more than one parameter
814 on platforms supporting dynamic-extent allocation.
815 * enhancement: saving cores with foreign code loaded is now supported
816 on MIPS/Linux in addition to the previously supported platforms.
817 * bug fix: FIND-RESTART now tests for activity, not applicability when given
818 a restart object as identifier. (reported by Helmut Eller for CMUCL)
819 * bug fix: division by zero in sb-sprof when no samples were collected
820 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
821 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
822 (thanks to Alexander Kjeldaas)
823 * new feature: ignoring the return values of destructive functions
824 like NREVERSE signals a compile-time style-warning.
825 (thanks to Kevin Reid)
826 * new feature: the alignment of alien structure fields can be
827 explicitly specified. (thanks to Cyrus Harmon)
828 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
829 and dump core on SIGQUIT
831 ** incompatible change: threads do not inherit values of specials
832 from their parents (see manual)
833 ** bug fix: threads stacks belonging to dead threads are freed by the
834 next exiting thread, no need to gc to collect thread stacks anymore
835 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
836 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
837 ** bug fix: dynamic variable and thread start related gc lossage
838 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
840 * fixed some bugs revealed by Paul Dietz' test suite:
841 ** SUBTYPEP is slightly more accurate on heinously complicated
842 CONS types where some of the members have uncertain (in the
843 NIL, NIL sense) type relationships to each other.
845 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
846 no longer purified unless :PURIFY T is explicitly specified.
847 ** Non-purified cores are significantly smaller than before
849 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
850 * new feature: timers based on Zach Beane's excellent timer package
851 * added support for the following external formats: koi8-u,
852 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
853 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
854 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
855 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
856 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
857 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
858 * incompatible change: a threaded SBCL will no longer revert to
859 non-threaded mode on non-NPTL systems, but refuse to start entirely.
860 * bug fix: interrupts are disabled until startup is complete; no
861 more sigsegvs when receiving a signal to soon
862 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
864 * bug fix: add a workaround for the memory randomization features in
865 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
866 workaround will only be in effect on systems with the proc filesystem
868 * bug fix: printing objects of type HASH-TABLE signals a
869 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
871 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
872 have been running for over 50 days. (reported by Gilbert Baumann)
873 * bug fix: the logic for getting names of functions gets less
874 confused when confronded with alternate-metaclass
875 funcallable-instances. (reported by Cyrus Harmon)
876 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
877 consistent, even on internal alternate-metaclass objects.
878 * bug fix: flush closure information collected by physical
879 environment analysis prepass before the main pass. (bug reported
881 * bug fix: compiler pack phase does not modify a hashtable, which is
882 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
883 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
884 may be expanded. (reported by Alexander <malishev> on c.l.l)
885 * bug fix: no more style-warnings when compiling code using LAST
886 under high-SPEED low-SPACE settings. (reported by David Wallin
888 * fixed some incompatibilities between SBCL's MOP and the MOP
890 ** the METAOBJECT class is now implemented;
891 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
892 STANDARD-OBJECT, as required;
893 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
894 now compatible (as required by VALIDATE-SUPERCLASS); there
895 remains a consistency requirement over the presence or absence
896 of the FUNCTION class in the superclass list of finalized
897 classes; see the manual for more details;
898 ** the :AROUND method for COMPUTE-SLOTS on
899 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
900 requested slot ordering.
902 ** bug fix: parent thread now can be gc'ed even with a live
904 ** bug fix: binding a special with PROGV to no value is not
905 the same as that symbol not having been bound (thanks to
907 ** bug fix: binding specials is thread safe (thanks to
909 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
910 and SB-SPROF do not die with 'no handler for signal XX in
911 interrupt_handle_now(..)' anymore
912 ** bug fix: WITH-TIMEOUT works with multiple running threads
913 * fixed some bugs revealed by Paul Dietz' test suite:
914 ** ENSURE-GENERIC-FUNCTION should take a method class object for
915 the :method-class keyword argument.
917 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
918 * new port: the Solaris operating system on x86 processors is now
919 mostly supported, though some rough edges in the environment
920 remain. (thanks to Daisuke Homma)
921 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
922 space for its heap. (thanks to Thiemo Seufer)
923 * optimization: SBCL on MIPS platforms now supports dynamic-extent
924 closures. (thanks to Thiemo Seufer)
925 * minor incompatible change: eof selects abort in the debugger.
926 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
927 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
929 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
930 deprecated, and will go away in a future revision of SBCL.
931 * minor incompatible change: GC-ON and GC-OFF are no longer
932 implemented with a counter, it does not matter how many times gc
933 is switched on or off
934 * bug fix: discriminating functions for generic function classes
935 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
936 make invalid assumptions about method precedence order. (reported
938 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
939 platforms. (thanks to Thiemo Seufer)
940 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
941 * bug fix: correct pathname printing with printer escaping is on.
942 (thanks to Kevin Reid)
943 * bug fix: complex VOP definitions in "user-space" no longer trigger
944 package locks. (reported by Zach Beane)
945 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
946 for funcallable-instance functions, and is no different from
948 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
949 characters in the current readtable. (reported by Nicholas Neuss)
950 * bug fix: SUBTYPEP on various CONS types returns more a more
951 accurate acknowledgment of its certainty.
952 * optimizations: REMOVE-DUPLICATES now runs in linear time on
953 lists in some cases. This partially fixes bug 384.
954 * flush all standard streams before prompting in the REPL and the
956 * bug fix: signal handling and triggering gc do not conflict
957 directly anymore, in particular a high frequency sb-sprof does
958 not prevent gc from running
959 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
960 approximation for timezone and DST information between the
961 universal time epoch and the smallest negative 32-bit time_t.
962 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
963 year 1899 so long as the encoded time is non-negative.
964 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
965 the correct answer for non-constant rotate amounts when performing
966 an inline 32-bit rotation.
968 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
969 there is only one thread in the session
970 ** bug fix: memory leak for streams created in one thread and
971 written to in another
972 ** bug fix: lockup when compiled with gcc4
973 ** bug fix: race that allows the gc to be triggered when gc is
975 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
976 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
978 ** bug fix: finished threads are gc'ed properly
979 * fixed some bugs revealed by Paul Dietz' test suite:
980 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
981 has arguments with a different set of applicable methods from
982 the orignal arguments.
983 ** The type error thrown by MAP now has a correct expected-type
985 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
986 name a compiled function.
987 ** Three MISC tests where a large bignum was improperly coerced to
988 a float that couldn't represent that bignum during type
989 derivation were fixed.
990 ** SUBTYPEP can now handle types involving the negation of a
991 list-form FUNCTION type.
992 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
993 as (COMPLEX (AND RATIO (NOT FIXNUM))).
994 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
996 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
997 * New feature: Experimental support for bivalent streams: streams
998 opened with :element-type :default now allow character and binary
999 (unsigned-byte 8) I/O
1000 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1001 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1003 * bug fix: no more highly sporadic "couldn't check whether ~S is
1004 readable" when reading a stream and an interrupt hits in the middle
1005 of a select system call
1006 * compiler better recognizes complex arrays (reported by Eduardo
1008 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1009 for variables on SBCL built with linkage-tables. (reported by Luis
1011 * various error reporting improvements.
1012 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1013 (Thanks to Thiemo Seufer)
1014 * bug fix: nested reader invokations work correctly
1015 * bug fix: it is possible to have simultaneous references to foreign
1016 code and foreign data with the same name.
1018 ** added x86-64 support
1019 ** incompatible change: the threading api now works with thread
1020 objects instead of thread ids
1021 ** bug fix: threads are protected from signals and interruption when
1022 starting up or going down
1023 ** bug fix: a race where an exiting thread could lose its stack to gc
1024 ** bug fix: don't halt on infinite error in threads if possible
1025 ** fixed numerous gc deadlocks introduced in the pthread merge
1026 ** bug fix: fixed thread safety issues in read and print
1027 ** bug fix: debugger doesn't hang on session lock if interrupted at
1028 an inappropriate moment
1029 ** bug fix: run-program is now thread safe(r)
1030 ** bug fix: inner with-recursive-lock no longer releases the mutex
1031 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1032 * fixed some bugs revealed by Paul Dietz' test suite:
1033 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1034 EXPECTED-TYPE slots filled.
1035 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1037 ** MAKE-CONDITION accepts classes as type-designators.
1038 ** COMPILE may never return NIL.
1039 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1040 range before calling Unix time functions
1042 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1043 * numerous signal handling fixes to increase stability
1044 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1045 * minor incompatible change: we now correctly canonize default
1046 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1047 as per the MOP, rather than the historical (INITARG INITFUNCTION
1048 INITFORM). (reported by Bruno Haible)
1049 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1050 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1051 for more information.
1052 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1053 pathname is a directory pathname.
1054 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1055 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1057 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1058 :METACLASS argument in addition to a class name. (reported by
1059 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1060 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1061 (reported by Bruno Haible; patch from Gerd Moellmann)
1062 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1064 * bug fix: more cleanups to the floating point exception handling on
1065 x86-64 (thanks to James Knight)
1066 * bug fix: the compiler does not try to constant fold calls of
1067 COERCE and BIT-* functions when they return freshly constructed
1068 objects. (reported by Nikodemus Siivola and Paul Dietz)
1069 * optimization: improved the MIPS versions of generic (in the
1070 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1071 * optimization: direct conversion of (unsigned-byte 32) to floats on
1072 the PowerPC platform.
1073 * optimization: structure instances with raw slots now use less
1074 memory, and probably show better memory locality. (thanks to
1076 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1077 are treated as requests for stack allocation on the x86-64,
1078 Alpha-32, PPC and SPARC platforms.
1079 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1080 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1081 (thanks to Tony Martinez)
1083 ** gcing a dead thread can no longer lead to lockups
1084 ** threads block signals until they are set up properly
1085 ** errno is no longer shared by threads
1086 ** interrupt-thread restores the eflags register on x86
1087 ** fixed some lockups due to gc/thread interaction
1088 * fixed some bugs revealed by Paul Dietz' test suite:
1089 ** invalid dotted lists no longer raise a read error when
1090 *READ-SUPPRESS* is T
1091 ** signal an error if a symbol that names a declaration is used
1092 as the name of a type, or vice versa
1093 ** allow using the (declare (typespec var*)) abbreviation for
1094 (declare (type typespec var*)) with all type specifiers
1095 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1096 ** always return NIL from PPRINT-POP when OBJECT is NIL
1097 ** don't signal errors when pretty-printing malformed LABELS,
1098 FLET or MACROLET forms
1099 ** declarations in a DOLIST body are also in scope for the
1101 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1103 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1106 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1107 * fixed cross-compiler leakages that prevented building a 32-bit
1108 target with a 64-bit host compiler.
1109 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1110 opened with :IF-EXISTS :OVERWRITE.
1111 * fixed bug 281: error for an invalid qualifier in a short-form method
1112 combination method is not signalled until the faulty method is called.
1113 * bug fix: iteration variable type inferrer failed to deal with open
1114 intervals. (reported by Alan Shields)
1115 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1116 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1117 or more runtime options were provided to the sbcl binary.
1118 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1119 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1120 * optimization: there's now a fast-path for fixnum arguments in the
1121 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1122 * optimization: the code generated on x86-64 is more compact thanks
1123 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1124 * bug fix: floating-point exception handling now partly works on
1125 x86-64. (Thanks to James Knight)
1126 * improvement to the MIPS backend from Thiemo Seufer: C-style
1127 64-bit long long arguments and return values to alien functions
1129 * fixed some bugs revealed by Paul Dietz' test suite:
1130 ** the type-error signalled from WARN has a filled-in DATUM slot.
1131 ** the type-error required when a stream is not associated with
1132 a file has the stream as its datum.
1133 ** type-errors on single-floats on x86-64 no longer have
1134 :INVALID-OBJECT as the datum
1135 ** the type-errors signalled for invalid function names now have
1136 a correct expected type
1137 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1138 for structures defined with a :TYPE. Documentation strings for
1139 typed structures are no longer immediately discarded
1140 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1141 broadcast streams and synonym streams. FILE-LENGTH now also works
1142 on broadcast streams.
1144 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1145 * incompatible change: the --noprogrammer option, deprecated since
1146 version 0.7.5, has been removed. Please use the equivalent
1147 --disable-debugger option instead.
1148 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1150 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1151 inoperational for a while now) has been completely removed.
1152 * null lexical environments are now printed as #<NULL-LEXENV>,
1153 significantly reducing the amount of clutter in typical backtraces.
1154 * documentation on weak pointers, finalization, and after GC hooks
1155 has been added to the manual.
1156 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1157 as other specialized array types, is much faster. SUBSEQ and
1158 COPY-SEQ on such arrays have also been sped up.
1159 * optimization: EQL is now more efficient when at least other argument
1160 is known to be of type (OR FIXNUM (NOT NUMBER)).
1161 * fixed bug: the runtime is now less vulnerable to changes in the
1162 size of the SBCL object on OS X, and virtual memory is reserved for
1163 all spaces that need to be at a fixed address.
1164 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1165 * fixed bug: finalizers and after GC hooks that cause consing are now
1167 * fixed bug: compiler error messages and summaries are now printed to
1168 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1169 * fixed inference of the upper bound of an iteration variable.
1170 (reported by Rajat Datta).
1171 * fixed bug 211e: calling local functions with duplicated constant
1172 keyword argument no longer causes a bogus style warning about an
1174 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1175 ftype declaration to be disregarded. (reported by Dave Roberts)
1176 * fixed bug 373: caused by erronous compilation of references to alien
1177 variables in the runtime on ppc/darwin.
1178 * fixed bug 376: CONJUGATE type deriver.
1179 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1180 (reported by Baughn on #lisp)
1181 * a cross-compiler bug on non-x86 platforms has been identified and
1182 fixed. (thanks to Bruno Haible)
1183 * improvements to the MIPS runtime code for increased stability.
1184 (thanks to Thiemo Seufer)
1185 * increased the maximimum compact environment size to allow
1186 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1187 * partial workaround for bug 135: don't name the function we're
1188 calling for hairy cases of EVAL, so as not to accumulate one environment
1189 entry per call to EVAL. (reported by Kevin Reid)
1190 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1191 * optimization: type testing for non-vector arrays should be faster.
1192 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1194 * bug fix: setting 31st element of a bit vector to zero did not work
1196 * bug fix: redefining a class definition which failed due to a
1197 previous accessor / function clash now works (but see BUGS entry
1198 #380 for more problems in this area). (thanks to Zach Beane)
1199 * the long form of DEFINE-METHOD-COMBINATION disables method group
1200 checking when given a single method group with pattern *. (thanks
1202 * on x86 compiler supports stack allocation of results of simple
1203 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1204 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1205 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1206 * fixed some bugs related to Unicode integration:
1207 ** the restarts for recovering from input and output encoding
1208 errors only appear when there is in fact such an error to
1210 * fixed some bugs revealed by Paul Dietz' test suite:
1211 ** MISC.549 and similar: late transformation of unsafe type
1212 assertions into derived types caused unexpected code
1214 ** SCALE-FLOAT type deriver is less wrong.
1215 ** type derivers for EXP, LOG and similar functions compute result
1216 types for complex arguments better.
1217 ** (MISC.563) CONJUGATE type deriver works for very restricted
1219 ** out-of-line type testers for character strings are available.
1220 ** EQUAL compiler transform understands specialness of objects
1222 ** accessing double-floats stored on the stack now works on x86-64.
1223 ** debugger internals could sometimes create invalid lispobjs,
1224 resulting in GC crashes.
1225 ** MISC.548: type check weakening can convert required type into
1227 ** initialization forms for bindings are not in scope of free special
1230 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1231 * incompatible change: thread support for non-NPTL systems has
1232 been removed - locking is buggy and unreliable. A threaded
1233 SBCL build will now warn at startup and refuse to create
1234 new threads, unless futex support is detected in the kernel
1235 * incompatible change: the top level REPL now has only an ABORT
1236 restart associated with it, not TOPLEVEL and ABORT as it used to.
1237 TOP and TOPLEVEL are now available as debugger commands for
1238 returning to the top level.
1239 * incompatible change: forms evaluated in the REPL now use the
1240 global optimization policy.
1241 * incompatible change: user- and system-initialization files are
1242 no longer processed with LOAD, but by READ and EVAL; hence the
1243 global optimization policy, startup package, readtable, etc,
1245 * The .fasl file format number has been incremented because of
1246 various incompatible changes.
1247 * internal entry point details and argument counts no longer appear
1248 in backtraces unless explicitly requested by setting
1249 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1250 * built-in and standard functions no longer have names like "top
1251 level local call to FOO".
1252 * fixed bug 32: functions defined in non-null lexical environments
1253 now have more legible printed representation
1254 * fixed bug 33: functions defined in non-null lexical environemnts
1255 are now more amenable to inspection by INSPECT.
1256 * workaround for bug 354: XEPs no longer appear in backtraces unless
1257 explicitly requested.
1258 * fixed bug: receiving the signal which results from attempting to
1259 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1260 SIGBUS on other *BSDs) on architectures where the C stack is also
1261 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1262 to debugger confusion or wild instability. (reported by Cheuksan
1264 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1265 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1266 (reported by Lutz Euler)
1267 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1268 compiled code. (reported by Paul Dietz)
1269 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1270 specialized parameter to a method to something that is not TYPEP
1271 the specializer is now possible.
1272 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1273 face of package deletion.
1274 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1275 pathname. (reported by David Tolpin on c.l.l)
1276 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1277 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1278 STRING-STREAM and FILE-STREAM, respectively.
1279 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1280 than previously; in addition, its test suite is now run on build.
1281 (thanks to Luke Gorrie)
1282 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1283 * optimization: added a immediate representation for single-floats
1285 * fixed some bugs revealed by Paul Dietz' test suite:
1286 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1287 ** attempting to create a package with a colliding nickname causes
1288 correctable errors to be signalled.
1289 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1290 ** several x86-64 backend bugs related to sign-extension of immediate
1293 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1294 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1295 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1297 * fixed loading of multiply forward-referenced layouts.
1298 (thanks to Cheuksan Wang)
1299 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1300 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1301 (see COMPILE-FILE documentation for details of :PRINT option)
1302 * fixed bugs 19 and 317: fixed-format floating point printing is
1303 more accurate. This also fixes a bug reported by Adam Warner
1304 related to the ~@F format directive.
1305 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1307 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1308 dispatch table if the from-char is a dispatch macro character.
1309 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1310 by the wordsize no longer ignore the last word. (reported by Lutz
1312 * fixed bug in type checking of dynamic-extent variables. (reported
1314 * optimization: sequence traversal functions use their freedom to
1315 coerce function designators to functions.
1316 * optimization: code with many calls to CLOS methods specialized on
1317 CLOS classes has had redundant type checks removed.
1318 * optimization: added declarations to speed up operations that access
1319 the internal character database (for example STRING-UPCASE)
1320 * optimization: comparison operations between floats and sufficiently small
1321 fixnums no longer create extra rationals
1322 * fixed some bugs related to Unicode integration:
1323 ** portions of multibyte characters at the end of buffers for
1324 character-based file input are correctly transferred to the
1325 start of the buffer at the next read.
1326 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1327 passing it through to OPEN.
1328 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1329 argument given to internal calls to OPEN.
1330 ** the built-in comment readers (introduced by character sequences
1331 ";" and "#|") are more forgiving to encoding errors; they will
1332 STYLE-WARN and then attempt to resync the stream at a character
1333 boundary. (thanks to Teemu Kalvas)
1334 * fixed some bugs in the x86-64 port:
1335 ** Negative short int return values from c-calls are sign-extended
1337 ** The stack is aligned to 16-bytes for c-calls, as required by
1338 the ABI. (thanks to Cheuksan Wang)
1339 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1340 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1341 for enums. (thanks to Vincent Arkesteijn)
1342 ** Multiple small optimizations and bugfixes for floating point
1344 * fixed some bugs revealed by Paul Dietz' test suite:
1345 ** Space, Tab, Linefeed, Return and Page have the invalid
1346 secondary constituent character trait.
1347 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1349 ** WITH-INPUT-FROM-STRING should only update the index place on
1351 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1352 works more reliably.
1353 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1354 with one-dimensional arrays with fill pointers.
1355 ** TYPE-OF failed on a complex with an integer realpart and a
1357 ** compiler failure during type inference for the code of form
1358 (IF (EQL X (THE ...)) ...) (MISC.535).
1360 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1361 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1362 platform. The port supports 61-bit fixnums, large memory spaces
1363 and reloading shared object files.
1364 * enhancement: saving cores with foreign code loaded is now
1365 supported on ppc/Darwin in addition to the previously supported
1367 * enhancement: the statistical profiler now walks deeper into the
1368 call stack for more meaningful call-graphs and accrued time
1369 reports (x86/x86-64 only). It also now reports time spent in
1371 * enhancement: it is now possible to trace most individual methods
1372 of a generic function in addition to tracing the generic function
1374 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1376 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1377 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1378 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1379 produces an error. (thanks to Vincent Arkesteijn)
1380 * bug fix: NAMESTRING on pathnames with :WILD components in their
1381 directories works correctly. (thanks to Artem V. Andreev)
1382 * fixed bug 125: compiler preserves identity of closures. (reported
1384 * bug fixed: functions with &REST arguments sometimes failed with
1385 "Undefined function" when compiled with (DEBUG 3). (reported by
1387 * bug fix: overflow during compiling of setting element of a bit
1388 vector with constant index and value. (reported by Timmy Douglas)
1389 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1390 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1391 * on x86 compiler partially supports stack allocation of dynamic-extent
1393 * GO and RETURN-FROM do not check the extent of their exit points
1394 when compiled with SAFETY 0.
1395 * fixed some bugs related to Unicode integration:
1396 ** encoding and decoding errors are now much more robustly
1397 handled; it should now be possible to recover even from invalid
1398 input or output to the terminal. (thanks to Teemu Kalvas)
1399 ** provided a first cut at implementing STRING-TO-OCTETS and
1400 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1401 ** altered the SB-MD5 contributed module slightly, changing the
1402 interface just enough for it to be supportable for builds where
1403 lisp characters are not eight bits.
1404 * fixed some bugs revealed by Paul Dietz' test suite:
1405 ** the FORMATTER-generated functions for ~V[ conditionals require
1406 the correct number of arguments.
1407 ** READ-FROM-STRING returns the mandated second value when applied
1408 to displaced strings.
1409 ** the #\Rubout and #\Backspace characters are treated as invalid
1410 constituent characters by the tokenizer.
1412 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1413 * new feature: reloading changed shared object files with
1414 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1415 * new feature: references to foreign variables and functions
1416 can now be compiled and loaded before the corresponding shared
1417 object file is loaded, as long as the foreign definitions are
1418 available at runtime.
1419 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1420 supported. (thanks to Dan Debertin)
1421 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1422 just keywords. (thanks to Vincent Arkesteijn)
1423 * fixed bug #331: structure-class instances corresponding to
1424 DEFSTRUCT forms are now created eagerly.
1425 * fixed bug #345: backtraces from calls to undefined functions work
1426 on x86 as well. Related bug #61 is now also partially fixed on x86
1427 (backtraces from throws to unknown catch tags.)
1428 * bug fix: lambda-list parsing is now stricter vrt. order and number
1429 of lambda-list keywords.
1430 * bug fix: as specified by AMOP, an error is signalled if a
1431 class-option appears multiple times in a DEFCLASS form. (reported
1433 * bug fix: RANDOM can be compiled when the compiler derives the type
1434 of its numeric argument as a disjoint set of small integers.
1435 (reported by Paul Dietz)
1436 * bug fix: starting a core saved with shared objects loaded when
1437 those objects are not available no longer causes threaded SBCL to
1438 hang. (reported by Sean Ross)
1439 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1441 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1442 argument, which must always be NIL. (reported by Kalle Niemitalo)
1443 * bug fix: printing 1.0d+23 no longer results in an error.
1444 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1445 * bug fix: structure slot setters preserve evaluation order. (thanks
1447 * fixed some bugs related to Unicode integration:
1448 ** RUN-PROGRAM can allow its child to take input from a Lisp
1449 stream. (reported by Stefan Scholl)
1450 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1451 locales. (reported by Ken Causey)
1452 * fixed some bugs revealed by Paul Dietz' test suite:
1453 ** INCF, DECF and REMF evaluate their place form as specified in
1455 ** FORMATTER expands ~{ ~} iteration directives with V or #
1456 parameters correctly.
1457 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1458 consequent uses no arguments correctly.
1459 ** the system has a partial understanding of the (COMPLEX RATIO)
1461 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1464 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1465 * new feature: a build-time option (controlled by the :SB-UNICODE
1466 keyword feature, enabled by default) for building the system with
1467 support for the entire 21-bit character space defined by the
1469 * new feature: the system now has rudimentary external-format
1470 support; the primary user-visible change at this time is that
1471 characters with the high bit set (such as lower-case-e-acute) will
1472 print correctly to a terminal in a UTF-8 environment.
1473 * minor incompatible change: BASE-CHAR no longer names a class;
1474 however, CHARACTER continues to do so, as required by ANSI.
1475 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1476 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1477 has been moved to the SB-EXT package (temporarily re-exported from
1479 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1480 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1482 * fixed bug #351: better error-handlind and reporting for malformed
1484 * fixed bug #350: bignum-printing is now more memory-efficient,
1485 allowing printing of very large bignums, eg. (expt 2 10000000).
1486 (reported by Bruno Haible)
1487 * fixed bug #302: better primitive-type selection for intersection
1489 * fixed bug #308: non-graphic characters now all have names, as
1490 required. (reported by Bruno Haible)
1491 * bug fix: redefining a class with different superclasses now correctly
1492 removes it from the direct-subclasses of its previous superclasses.
1493 (reported by David Morse)
1494 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1495 new value now works. (reported by Bruno Haible)
1496 * bug fix: correct canonicalization of multiple non-standard slot
1497 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1498 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1499 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1500 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1501 now exists, an signals an error.
1502 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1503 during macroexpansion for non-iteration variables as well. (reported
1504 by Bruno Haible for CMUCL)
1505 * bug fix: Cyclic structures and unprintable objects in compiler
1506 messages no longer cause errors. (reported by Bruno Haible)
1507 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1508 and READ-FROM-STRING all now return a primary value of NIL if
1509 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1510 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1511 (reported by Bruno Haible for CMUCL)
1512 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1513 do not designate a condition. (reported by Bruno Haible for
1515 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1516 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1517 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1518 permitting resolution in favour of any of the conflicting symbols.
1519 (reported by Bruno Haible for CMUCL)
1520 * bug fix: EQUAL compiler optimizations is less aggressive on
1521 strings which can potentially compare true despite having distinct
1522 specialized array element types.
1523 * bug fix: unit enumerations can be defined without dividing by
1524 zero. (thanks to Vincent Arkesteijn)
1525 * FORMAT compile-time argument count checking has been enhanced.
1526 (report from Bruno Haible for CMUCL)
1527 * a partial workaround for the bug 262: the compiler does not try to
1528 inline-expand a local function doing RETURN-FROM from a deleted
1529 BLOCK. (thanks to Peter Denno for the bug report and to David
1530 Wragg for the simple test case)
1531 * fixed some bugs revealed by Paul Dietz' test suite:
1532 ** RENAME-PACKAGE allows all package designators as new package
1534 ** constraint propagation and lambda variable substitution are
1535 more cautious in dealing with partially deleted code.
1536 ** compiler sometimes lost reoptimization passes.
1537 ** CERROR, given a condition as condition designator, passes its
1538 remaining arguments to the continue format control without
1540 ** Case-altering FORMAT directives work correctly on non-ASCII
1542 ** The REAL type specifier handles bounds outside the floating
1543 point ranges without signalling FLOATING-POINT-OVERFLOW.
1544 ** Functions with IR1-transformations can create intercomponent
1545 references to global functions.
1546 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1548 ** FORMAT ~R treats a nil value for its first parameter correctly.
1550 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1551 * enhancement: saving cores with foreign code loaded is now
1552 supported on x86/NetBSD and sparc/Linux in addition to the previously
1553 supported platforms.
1554 * bug fix: on some platforms repeated installations caused multiple
1555 copies of HTML documentation to be installed -- should not happen
1556 any more. (reported by Stefan Scholl)
1557 * bug fix: parsing self-recursive alien record types multiple times
1558 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1559 original patch by Helmut Eller for CMUCL)
1560 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1561 (thanks to Richard Kreuter)
1562 * bug fix: defining classes whose accessors are methods on existing
1563 generic functions in other (locked) packages no longer signals
1564 bogus package lock violations. (reported by François-René Rideau)
1565 * bug fix: special variables as DEFMETHOD parameters no longer have
1566 associated bogus type declarations. (reported by David Wragg and
1568 * bug fix: read-write consistency on streams of element-type
1569 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1570 * bug fix: redefiniton of the only method of a generic function with
1571 no DEFGENERIC no longer emits a full WARNING. In addition,
1572 redefinition of generic functions with no DEFGENERIC to an
1573 incompatible lambda list now signals an error. (thanks to Zach
1575 * bug fix: DEFGENERIC now works even when there's a function of the
1576 same name in an enclosing lexical environment. (thanks to Zach
1578 * fixed compiler failure, caused by instrumenting code during
1579 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1580 * optimization: added loop analysis and improved register allocation
1581 * fixed some bugs revealed by Paul Dietz' test suite:
1582 ** POSITION on displaced vectors with non-zero displacement
1583 returns the right answer.
1584 ** (SIMPLE-STRING) is a valid type specifier for sequence
1586 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1588 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1589 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1591 ** The justification version of the FORMAT ~< directive treats
1592 non-zero minpad parameter correctly.
1594 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1595 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1596 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1597 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1598 the supported interface.
1599 * new feature: Single-stepping of code compiled with DEBUG 2 or
1600 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1601 * new feature: saving cores with foreign code loaded is now
1602 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1603 Timothy Moore's work for CMUCL)
1604 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1605 parameters to * instead of NIL if no initform is supplied.
1606 (reported by Johan Bockgaard)
1607 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1608 correctly when FUNCALL appears as the car of the form. Note:
1609 despite this FUNCALL forms are not currently subject to
1610 compiler-macro expansion. (port of Raymond Toy's fix for the
1611 same from CMUCL, reported by Johan Bockgaard)
1612 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1613 (thanks for Teemu Kalvas)
1614 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1615 case instead of terminating the iteration (thanks for Julian
1616 Squires, Sean Champ and Raymond Toy)
1617 * bug fix: incorrect expansion of defgeneric that caused a style
1618 warning. (thanks for Zach Beane)
1619 * bug fix: slot accessor effective method computation works properly
1620 for classes with multiple non-standard applicable methods on
1621 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1622 * on x86 compiler supports stack allocation of results of LIST and
1623 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1624 CMUCL implementation by Gerd Moellmann)
1625 * fixed some bugs revealed by Paul Dietz' test suite:
1626 ** FORMAT strings with both the ~<~:;~> form of the justification
1627 directive and pretty-printing directives cause an error.
1629 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1630 * incompatible change: the internal functions
1631 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1632 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1633 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1634 instead of the old functions.
1635 * new feature: on platforms where "dladdr" is available foreign
1636 function names now appear in backtraces. (based on Helmut Eller's
1638 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1639 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1641 * documentation: ASDF manual is now distributed with SBCL.
1642 * bug fix: SBCL can now load its contributed modules with REQUIRE
1643 even if the system-provided entries have been removed from
1644 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1646 * bug fix: backtraces involving undefined functions or assembly
1647 routines are more informative. (thanks to Brian Downing)
1648 * bug fix: mutually referent alien structures now work correctly.
1649 (reported by Rick Taube)
1650 * bug fix: structures defined by WITH-ALIEN can be referred to
1651 within other definitions in the same WITH-ALIEN.
1652 * bug fix: division operators (MOD, TRUNCATE and the like) with
1653 constant zero divisors and integer dividends no longer generate
1655 * bug fix: provide default methods for INPUT-STREAM-P and
1656 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1657 * bug fix: improve the detection and resolution of MOP metacycles
1658 (where effective method computation depends on the generic function
1659 having its effective method computed). (reported by Bruno Haible)
1660 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1661 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1662 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1663 * bug fix: pathnames with relative directory components can be
1664 represented relative to default pathnames.
1665 * optimization: in taking the GCD of bignums, reduce the two bignums
1666 to approximately the same size (using Euclid's algorithm) before
1667 applying the more sophisticated binary GCD. (thanks to Juho
1669 * optimization: COUNT on bitvectors now operates word-at-a-time.
1670 * optimization: ASH with a positive, but not necessarily constant,
1671 (leftwards) shift, when in a modular context, is compiled to a
1673 * fixed some bugs revealed by Paul Dietz' test suite:
1674 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1675 if the corresponding argument is NIL.
1676 ** FORMAT directives accepting a minpad parameter treat negative
1678 ** literal commas as character parameters to FORMAT directives are
1680 ** literal spaces directly after ~<Newline> directives within a
1681 format-logical-block (~:< ~@:>) do not induce :FILL-style
1682 conditional newlines.
1683 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1684 as specified: it no longer includes conditional newlines).
1685 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1687 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1688 :PER-LINE-PREFIX argument does not evaluate to a string.
1690 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1691 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1692 the manual for details; for now, package locks can be disabled by
1693 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1694 if no major problems are found then it is likely that they will be
1695 compiled in unconditionally.
1696 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1697 unsupported operators on all platforms. To load a shared library
1698 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1699 object file, link it into a shared library outside of SBCL and
1700 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1702 * bug workaround: ROOM T and the GC were not working together
1703 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1704 reporting facility aren't preserved. That reporting has been
1705 disabled, so now until and unless someone figures out how to make it
1706 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1707 * minor incompatible change: as threatened around sbcl-0.8.0, the
1708 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1709 The symbols are also exported from SB-PCL for backwards
1710 compatibility, but more so than before SB-PCL should be treated as
1711 an implementation-internal package.
1712 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1714 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1715 It is known as of this release not to work on the Alpha, however.
1716 * fixed bug #167: errors signalled due to illegal syntax in method
1717 bodies are now more legible.
1718 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1719 designators and can hence be used with TYPEP.
1720 * fixed bug #333: CHECK-TYPE now ensures that the type error
1721 signalled, if any, has the right object to be accessed by
1722 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1723 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1724 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1726 * fixed bug #334: programmatic addition of slots using specialized
1727 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1728 and :ALLOCATION :CLASS slots.
1729 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1730 float exponents. (rereported by Peter Seibel)
1731 * fixed a bug: #\Space (and other whitespace characters) are no
1732 longer considered to be macro characters in standard syntax by
1733 GET-MACRO-CHARACTER.
1734 * fixed bug: initialization of condition class metaobjects no longer
1735 causes an instance of the condition to be created. (reported by
1737 * fixed bug: it is now possible to have slots such that
1738 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1739 description is neither :INSTANCE nor :CLASS.
1740 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1741 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1742 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1743 non-standard SLOT-DEFINITION-ALLOCATION values.
1744 * fixed bugs: various race conditions handling exiting threads.
1745 CL-PPCRE's thread torture test now passes.
1746 * fixed arguably-a-bug: GC time proportional to number of threads in
1747 system even when most of them are idle
1748 * optimization: improved performance of BIT and SBIT on bit-vectors.
1749 * .fasl file incompatibility: The fasl file version number has
1750 been incremented because of changes associated with package locks.
1752 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1753 * minor incompatible change: the system no longer provides
1754 optimization hints (conditions which are TYPEP
1755 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1756 mode; these hints are still emitted when higher SPEED optimization
1758 * new contrib module: a sampling profiler (profiling by statistical
1759 sampling, rather than by instrumenting functions) is available as
1760 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1761 * the behaviour of the standard function ED is now customizeable by
1762 third parties through a hook variable: see ED's documentation
1763 string for information on the protocol.
1764 * the compiler no longer emits efficiency notes for (FUNCALL X)
1765 when the type of X is uncertain under default optimization
1767 * fixed bug 276: mutating a binding of a specialized parameter to a
1768 method to something that is not TYPEP the specializer is now
1770 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1771 exhibit the required behaviour on the x86 platform. (thanks to
1772 Peter van Eynde, Eric Marsden and Bruno Haible)
1773 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1774 even for difficult arguments. (reported by Peter Graves)
1775 * fixed bug 141a: the backquote printer now descends quoted
1777 * fixed another bug in backquote printing: no more destructive
1778 modification of the form's list structure. (reported by Brian
1780 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1781 move between its address being taken and the call to
1782 interrupt_thread, fixing a crashing race condition.
1783 * the SB-POSIX contrib implementation has been adjusted so that it
1784 no longer exhibits ridiculously poor performance when constructing
1785 instances corresponding to C structs.
1787 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1788 * minor incompatible change: the sb-grovel contrib now treats C
1789 structures as alien (in the sense of SB-ALIEN) objects rather than
1790 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1791 has implications for memory management of client code
1792 (sb-grovel-returned objects must now be manually managed) and for
1793 type safety (alien objects now have full types).
1794 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1795 used to control emission of compiler diagnostics, rather than the
1796 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1797 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1798 quality should be considered deprecated.
1799 * install.sh now installs the user manual as well
1800 * (not quite a new documentable feature, but worth considering in
1801 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1802 beginnings of a semantically meaningful condition hierarchy is
1803 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1805 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1806 designator as the defaults argument.
1807 * fixed bug: Displaced arrays whose displaced-to array has become
1808 too small now cause ARRAY-DIMENSION to signal an error, providing
1809 for safer bounds-checking. (reported by Bruno Haible)
1810 * fixed bug: DEFCLASS slot definitions with identical :READER and
1811 :WRITER names now signal a reasonable error. (reported by Thomas
1813 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1815 * fixed bug: Invalid slot specification errors now print correctly.
1816 (thanks to Zach Beane)
1817 * fixed bug 320: Shared to local slot value transfers in class
1818 redefinitions now happen corrently from superclasses as
1819 well. (reported by Bruno Haible)
1820 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1822 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1823 improper lists correctly. (reported by Bruno Haible)
1824 * fixed bug 313: source-transform for <fun-name> was erroneously
1825 applied to a call of a value of a variable with name <fun-name>.
1826 (reported by Antonio Menezes Leitao)
1827 * fixed bug 307: The obsolete instance protocol ensures that
1828 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1829 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1830 remove dead unknown-values globs from the middle of the stack.
1831 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1832 easier to resume long computations after using *BREAK-ON-SIGNALS*
1833 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1834 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1835 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1836 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1837 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1838 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1840 * fixed a (fairly theoretical) bug in string printing: if
1841 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1842 does not have array-element-type equal to the most general string
1844 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1845 function name. (thanks to Nikodemus Siivola)
1846 * fixed bug: compile-time format string checker failed on
1847 non-closed ~{. (reported by Thomas F Burdick)
1848 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1849 don't warn on legal loop code involving more than one aggregate
1850 boolean. (thanks to Nikodemus Siivola)
1851 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1852 in keyword position to MAKE-PACKAGE was not being reported as
1854 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1855 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1857 * fixed bugs 280 and 312: the checking for multiple definitions in a
1858 file is less likely to become confused by uses of inline
1860 * fixed bug: the #S reader macro performs the keyword coercion
1861 specified for slot names. (reported by Kalle Niemitalo)
1862 * fixed bug: lambda lists may contain symbols whose names start with
1863 & but are not lambda-list-keywords; their occurrence triggers a
1865 * fixed bug 321: define-method-combination argument lambda lists do
1866 not cause invalid code to be generated when &optional or &aux
1867 variables have default values. (reported by Bruno Haible)
1868 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1869 have CLOS classes; this appears also to have fixed reported
1870 instability in stack exhaustion detection.
1871 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1872 MUFFLE-WARNING when no associated restart is present is now
1874 * optimization: rearranged the expansion of various defining macros
1875 so that each expands into only one top-level form in a
1876 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1878 * optimization: used a previously little-used slot in symbols to
1879 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1881 * fixed some bugs revealed by Paul Dietz' test suite:
1882 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1884 ** Fixed an optimization bug related to inheritance of initforms
1885 from local to shared slots.
1886 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1887 ** CAST optimizer forgot to flush argument derived type.
1888 ** print/read consistency on floats is now orders of magnitude
1889 more likely. (thanks also to Bruno Haible for a similar report
1891 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1892 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1893 for objects of type REAL. Make it so.
1894 ** SXHASH is specified (infelicitously) to respect similarity,
1895 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1896 it so. (thanks to Markus Ziegler)
1897 ** on the Alpha, the compiler succeeds in compiling functions
1898 returning a known number of arguments greater than 63.
1899 ** fixed handling of invalid NIL arguments in keyword position
1901 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1903 ** provided a readably-printable representation for RANDOM-STATE
1905 ** ensured that pathnames, where they have a namestring, always
1906 print using #P"..." syntax.
1908 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1909 * Support for the forthcoming 2.0 version of the NetBSD kernel
1910 running on x86 hardware has been added. (thanks to Perry
1911 E. Metzger most immediately, and others for their past work)
1912 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1913 current release version; SBCL's OpenBSD support had been broken
1914 since about the time of OpenBSD's switch to ELF binary format.)
1915 * [placeholder for DX summary]
1916 ** user code with &REST lists declared dynamic-extent, under high
1917 speed or space and low safety and debug optimization policy.
1918 * The manual has been converted to Texinfo format and the debugger
1919 chapter from the cmucl manual has been added.
1920 * A facility has been added to extract documentation strings from
1921 sbcl and store them as Texinfo-formatted snippets for inclusion in
1922 the manual (via Texinfo's @include directive)
1923 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1924 the test case to Dave Roberts)
1925 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1926 pointers. (reported by Sean Ross)
1927 * bug fix: PROFILE output is printed nicely even for large numerical
1928 values. (thanks to Zach Beane)
1929 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1930 greater than 32 handle EOF correctly.
1931 * bug fix: on X86 an immediate argument of the IMUL instruction is
1932 correctly printed in disassembly. (thanks to Lutz Euler)
1933 * bug fix: class slots in redefined classes preserve their old
1934 values. (thanks to Bruno Haible and Nikodemus Siivola)
1935 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1936 succeeds. (reported by Marco Baringer)
1937 * fixed some bugs revealed by Paul Dietz' test suite:
1938 ** READ-SEQUENCE now works on ECHO-STREAMs.
1939 ** RATIONALIZE works more according to its specification. (thanks
1942 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1943 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1944 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1945 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1946 should matter to you only if you rebind the printer control
1947 variables and then find you want different bindings in the
1948 debugger than in the ordinary execution of your program.)
1949 * The runtime build system has been tweaked to support building
1950 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1951 assembler and linker. (thanks to Nikodemus Siivola)
1952 * Unbound, undefined, undeclared variables now trigger full
1953 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1954 more often programmer error than deliberate exploitation of undefined
1956 * optimization: the hash algorithm for strings has changed to one
1957 that is less vulnerable to spurious collisions. (thanks to Juho
1959 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1960 needless bounds checking. (thanks to Juho Snellman)
1961 * optimization: implemented multiplication as a modular
1962 (UNSIGNED-BYTE 32) operation on the PPC backend.
1963 * fixed some bugs revealed by Paul Dietz' test suite:
1964 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1966 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1967 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1968 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1970 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1972 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1974 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1975 their output stream on EOF from read.
1976 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1977 have been read to end-of-file.
1978 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1980 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1981 description of determination of which consecutive characters
1983 ** Printing the "Space" character with escaping on now yields "#\\ ",
1984 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1985 ** Reading floating-point numbers with *READ-BASE* set to a number
1986 less than 10 works correctly.
1987 ** Reading floating-point numbers with *READ-BASE* set to a number
1988 more than 10 works correctly.
1989 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1990 the readtable currently in effect.
1992 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1993 * minor incompatible change: parsing of namestrings on a physical
1994 (Unix) host has changed; numbers after the final #\. in a
1995 namestring are no longer interpreted as a version field. This is
1996 intented to be largely invisible to the user, except that the
1997 meaning of the namestring "*.*.*" has changed: it now refers to a
1998 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1999 should usually be replaced by
2000 (make-pathname :name :wild :type :wild :version :wild)
2001 with the added benefit that this is more likely to be portable.
2002 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2003 option to OPEN now signals an error if the file being opened
2004 exists; this may have an impact on existing code.
2005 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2006 signals received, on the PowerPC platforms (both Linux and
2007 Darwin). (thanks to Pierre Mai for pointing out the location of
2009 * several fixes on OS X: The system now builds and runs cleanly on
2010 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2011 (thanks to Brian Mastenbrook)
2012 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2013 second-resolution: integer multiples of 1/3600 between -24 and 24.
2014 (thanks to Vincent Arkesteijn)
2015 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2016 type when called with 1 argument; PEEK-CHAR checked type of
2017 PEEK-TYPE only after having read first character from a
2018 stream. (reported by Peter Graves)
2019 * bug fix: the garbage collector now has much better locality
2020 behaviour, and in particular no longer treats all memory as being
2021 exhausted when many small objects point to each other in a deeply
2023 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2024 recognized as being TYPEP their class.
2025 * bug fix: the PUSHNEW documentation string has been corrected.
2026 (thanks to Vincent Arkesteijn)
2027 * bug fix: defaulting of the value for the last of an atypically
2028 large number of multiple values being bound was not being
2029 performed correctly on the Alpha or PPC platforms
2030 * optimization: implemented multiplication as a modular
2031 (UNSIGNED-BYTE 32) operation on the x86 backend.
2032 * optimization: SEARCH on simple-base-strings can now be open-coded.
2033 (see also contrib/compiler-extras.lisp for inspiration for
2034 teaching the compiler about the Boyer-Moore algorithm).
2035 * value, returned by MAX (and MIN) called with several EQUALP, but
2036 not EQL, arguments now does not depend on compiler settings.
2037 * fixed some bugs revealed by Paul Dietz' test suite:
2038 ** in stack analysis liveness information is propagated from
2039 non-local entry points.
2040 ** pathwise CAST removing failed when the CAST node did not start
2042 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2043 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2045 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2046 host is already defined.
2047 ** RENAME-FILE works on streams instead of signalling an internal
2049 ** PEEK-CHAR uses the current readtable when determining whether
2050 or not a character is whitespace.
2051 ** MERGE-PATHNAMES handles the case when the pathname does not
2052 specify a name while the default-pathname specifies a version.
2053 ** Pathnames now stand a better chance of respecting print/read
2055 ** Attempting to use standardized file system operators with a
2056 pathname with invalid :DIRECTORY components signals a
2058 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2060 ** DIRECTORY on logical pathnames is more correct.
2061 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2062 signal a TYPE-ERROR if their argument is not a stream.
2063 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2064 designator argument does not designate a stream.
2065 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2066 examining the synonym.
2067 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2069 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2070 element-type larger than ([UN]SIGNED-BYTE 32).
2072 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2073 * When built with the :SB-FUTEX feature, threaded builds now take
2074 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2075 for faster/more reliable mutex and condition variable support.
2076 * Incompatible change (but one you probably shouldn't have been using
2077 anyway): the interface and code for arbitrating between multiple
2078 threads in the same user session has been redesigned.
2079 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2080 error when called without an explicit environment argument.
2081 (thanks to Nikodemus Siivola)
2082 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2083 confused about their position. (thanks to Adam Warner and Gerd
2085 * bug fix: Pretty printing backquoted forms with unquotations in the
2086 argument list position of various code constructs such as LAMBDA
2087 now works correctly. (reported by Paul Dietz)
2088 * bug fix: Pretty printing unquotations no longer loses all
2089 stream position information.
2090 * optimization: performance of string output streams is now less
2091 poor for multiple small sequence writes.
2092 * optimization: performance of CSUBTYPEP in the presence of complex
2093 expressions involving CONS and NOT many times has been improved.
2094 (reported by Paul Dietz)
2095 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2097 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2098 streams for the REPL, for the debugger, and so on. (thanks to
2100 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2101 * fixed some bugs revealed by Paul Dietz' test suite:
2102 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2103 is ignored if it is too small, rather than propagating through
2104 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2105 ** extremely complex negations of CONS types were not being
2106 sufficiently canonicalized, leading to inconsistencies in
2108 ** VALUES tranformer lost derived type.
2110 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2111 * fixed a bootstrapping bug: the build process no longer assumes
2112 that the various BOOLE-related constants have the same value in
2113 host and target lisps. (noted by Paul Dietz' test suite on an
2114 SBCL binary built from CLISP)
2115 * The system can now be dynamically linked on the MIPS platform,
2116 which enables dynamic loading of foreign code from Lisp. (thanks
2117 to Ralf Baechle for discussions on the MIPS ABI)
2118 * The system now records debugging information for its own source
2119 files in a filesystem-position-independent manner, relative to
2120 the "SYS" logical host.
2121 * fixed a compiler bug: MV-LET convertion did not check references
2122 to the "max args" entry point. (reported by Brian Downing)
2123 * tweaked disassembly notes to be less confident about proclaiming
2124 some instruction as an LRA. (thanks to Brian Downing)
2125 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2126 now each have their own history, command character, and other
2127 characteristics. (thanks to David Lichteblau)
2128 * fixed some bugs revealed by Paul Dietz' test suite:
2129 ** compiler failure in compiling LOGAND expressions including a
2131 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2132 shift greater than 32.
2133 ** FLUSH-DEST did not mark blocks for type check regeneration.
2134 ** HANDLER-CASE failed to accept declarations in handler clauses
2135 in some circumstances.
2137 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2138 * New code in contrib/sb-introspect (still probably not entirely
2139 stable yet) provides some support for smart Lisp development
2140 environments like SLIME.
2141 * The conditions signalled for errors occurring when loading .fasl
2142 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2143 in a way which should help ASDF recover gracefully.
2144 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2145 stuff has been cleaned up. If you code contrib/ stuff, this might
2146 affect you, and you can look at contrib/README, contrib/STANDARDS,
2147 and/or the 0.8.4.27 diff to check.
2148 * In full calls the compiler now does not generate checks for declared
2149 argument types for all arguments.
2150 * various threading fixes
2151 ** and some experimental patches which didn't make it into
2152 the main tree for this release, but which are shipped in
2153 contrib/experimental-thread.patch as a possible fix for some
2154 failures (deadlock, spinning...) in GC-intensive multithreaded
2156 * fixed PPC build problem (source code incompatibility of different
2157 library versions): added offsetof() hackery which attempts to divine
2158 where glibc maintainers put uc_mcontext today
2159 * fixed bug 282: compiler does not trust type assertions while passing
2160 arguments to a full call.
2161 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2162 &OPTIONAL VALUES type parameter.
2163 * fix bug 214: algorithm for noting rejected templates is now more
2164 similar to that of template seletion. (also reported by rydis on
2166 * fixed bug 141b: printing backquoted information readably and prettily
2167 inserts a space where necessary.
2168 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2169 (CHARACTER 10) are now reported as errors, rather than propagated
2170 as unknown types. (reported by piso on #lisp)
2171 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2172 with values NIL and :ERROR. (thanks to Milan Zamazal)
2173 * fixed bug 191c: CLOS now does proper keyword argument checking as
2174 described in CLHS 7.6.5 and 7.6.5.1.
2175 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2176 longer raise an error; further, using a list as a for-as-arithmetic
2177 counter now raises a meaningful error.
2178 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2179 now understood by sequence creation functions such as MAKE-SEQUENCE
2181 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2182 a STRING-INPUT-STREAM.
2183 * compiler enhancement: SIGNUM is now better able to derive the type
2185 * type declarations inside WITH-SLOTS are checked. (reported by
2187 * fixed some bugs revealed by Paul Dietz' test suite:
2188 ** incorrect optimization of TRUNCATE for a positive first
2189 argument and negative second.
2190 ** compiler failure in let-convertion during flushing dead code.
2191 ** compiler failure while deriving type of TRUNCATE on an
2192 interval, containing 0.
2193 ** ASH of a negative bignum by a negative bignum count now returns
2195 ** intersection of CONS types now canonicalizes properly, fixing
2196 inconsistencies in SUBTYPEP.
2198 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2199 * incompatible change: The --disable-debugger command line
2200 option now clobbers the debugger at a more fundamental
2201 level, by redefining #'INVOKE-DEBUGGER instead of by
2202 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2203 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2204 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2205 enter the debugger REPL and then suffer recursive errors
2206 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2207 while under the new system, BREAK in --disable-debugger mode
2208 terminates the system just as an unhandled error would.
2209 * fixed compiler performance when processing loops with a step >1;
2210 * bug fix: DOCUMENTATION now retrieves generic function
2211 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2212 support has been systematized, and now supports the methods
2213 specified by ANSI, along with a default method and a method for
2214 slot documentation. (reported by Nathan Froyd)
2215 * bug fix: effective methods associated with a generic function are
2216 no longer cached over a change of that generic function's method
2217 combination. (reported by Andreas Fuchs)
2218 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2219 names a non-existent directory. (thanks to Andreas Fuchs)
2220 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2221 the PPC platform, lead to stack corruption; this has been fixed.
2222 (reported by Rainer Joswig)
2223 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2224 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2225 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2226 arithmetic in general, is now much more reliable on the PPC
2228 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2229 platform now returns the right answer.
2230 * optimization: restored some effective method precomputation in
2231 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2232 precomputation is now tunable.
2233 * optimization: compiler-internal data structure use has been
2234 reviewed, and changes have been made that should improve the
2235 performance of the compiler by about 20%.
2236 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2237 simple-base-strings and simple-bit-vectors is improved.
2238 * optimization: the optimization of 32-bit logical and arithmetic
2239 functions introduced in version 0.8.3 on the x86 has been
2240 implemented on the mips, ppc and sparc platforms; an
2241 implementation of the same facility, but for 64-bit arithmetic,
2242 has been added for the alpha.
2243 * microoptimization: the compiler is better able to make use of the
2244 x86 LEA instruction for multiplication by constants.
2245 * bug fix: in some situations compiler did not report usage of
2246 generic arithmetic in (SPEED 3) policy.
2247 * bug 145b fix: compiler used wrong type specifier while converting
2248 MEMBER-types to numeric.
2249 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2251 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2252 index argument now works properly on the Alpha platform.
2253 * bug fix: floating point exception treatment on the Alpha platform
2255 * bug fix: FILE-POSITION works much better on string input and
2256 output streams. (thanks to Nikodemus Siivola)
2257 * bug fix: many threading/garbage collection symptoms sorted.
2258 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2259 might be pseudo-atomic.
2260 * internal change: Stopping for GC is now done with signals not
2261 ptrace. GC is now done in whichever thread wanted it, instead of
2263 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2265 * bug fix: result form in DO is not contained in the implicit
2267 * incompatible change: ICR structure is changed; the value part of
2268 CONTINUATION is now called LVAR; corresponding functions are
2269 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2270 * added type deriver for ISQRT (thanks to Robert E. Brown).
2271 * bug fix: better support for loading from the command line when an
2272 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2274 * fixed some bugs revealed by Paul Dietz' test suite:
2275 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2276 ** ROUND and FROUND now give the right answer when given very
2277 small float arguments.
2278 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2280 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2281 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2282 ** LCM should return a non-negative integer.
2283 ** PARSE-INTEGER returned the index of a terminator instead of the
2284 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2285 ** PARSE-INTEGER returned an incorrect index being applied to a
2287 ** LCM with two arguments of 0 returns 0 rather than signalling
2289 ** unsigned addition of a 32-bit constant with the high bit set no
2290 longer causes an internal compiler error on the x86.
2291 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2292 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2293 with negative last argument.
2294 ** byte specifiers with zero size and position no longer cause
2295 an error during type derivation.
2296 ** bignum multiplication on the Alpha platform now returns the
2298 * porting: The system now builds on SuSE AMD64, although it still
2299 generates a 32-bit binary.
2300 * .fasl file incompatibility: The fasl file version number has
2301 been incremented (because of the changes to internal compiler
2302 data structures referred to above).
2304 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2305 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2306 more accurately, on the Darwin kernel running on PowerPC hardware.
2307 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2308 * Compiler code deletion notes now signal a condition of type
2309 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2310 an associated MUFFLE-WARNING restart.
2311 * The compiler now performs limited argument count validation of
2312 constant format strings in FORMAT, and where appropriate in ERROR,
2313 CERROR and WARN. (thanks to Gerd Moellmann)
2314 * New ASDF-INSTALL contrib can be used for automatic download and
2315 installation of third-party Lisp code from CCLAN or other sites
2317 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2318 which forces another thread to execute a function supplied by the
2320 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2321 now accept and act upon their :ELEMENT-TYPE keyword argument.
2322 (reported by Martin Atzmueller, Edi Weitz)
2323 * bug fix: FILE-POSITION now accepts position designators up to
2324 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2325 is the greater. (thanks to Patrik Nordebo)
2326 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2327 Kalle Olavi Niemitalo)
2328 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2329 printable. (reported by Eric Marsden)
2330 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2331 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2332 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2333 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2334 (thanks to Stig Sandoe)
2335 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2336 UNREAD-CHAR now correctly includes the unread character in the
2337 target sequence. (thanks to Gerd Moellmann)
2338 * bug fix in threaded builds: the system can now be suspended and
2339 resumed by shell job control with minimal disruption.
2340 * bug fixes in times and timezones >2038AD
2341 * better handling of "where is GNU make?" problem in build scripts
2342 (thanks to Nikodemus Siivola)
2343 * new optimization: inside a named function any reference to a
2344 function with the same name is considered to be a self-reference;
2345 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2346 optimization quality.
2347 * new optimization on x86: logical functions and + now have
2348 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2349 used when the result is truncated to 32 bits.
2350 * VALUES declaration is partially enabled.
2351 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2352 * fixed some bugs revealed by Paul Dietz' test suite:
2353 ** The system now obeys the constraint imposed by
2354 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2355 types form a lattice under type intersection.
2356 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2357 ** ASSOC now ignores NIL elements in an alist.
2358 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2359 and (1+ MOST-POSITIVE-FIXNUM) answers.
2360 ** The addition of a method with invalid qualifiers to a generic
2361 function does not cause an error to be signalled immediately;
2362 a warning is signalled, and the error is generated only on
2363 calling the generic function.
2364 * changed .fasl file version number, in part to add type codes for
2365 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2366 obscure ANSI requirements
2368 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2369 * fixed bug 148: failure to inline-expand a local function left
2370 garbage, confusing the compiler.
2371 * fixed bugs 3cd: structure slot readers perform type check if the
2372 slot can have an invalid value (i.e. it is either not initialized
2373 or can be written with a less specific slot writer).
2374 * bug fix: the compiler now traps array references to elements off
2375 the end of an array; previously, the bounds checking in some
2376 circumstances could go off-by-one.
2377 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2379 * type declarations for array element types now obey the description
2380 on the CLHS page "Declaration TYPE", as per discussions on
2381 sbcl-help around 2003-05-08. This means that a declaration
2382 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2383 declaration, all references to BAR will be asserted or assumed
2384 (with THE, so dependent on compiler policy) to involve objects of
2385 type FOO. Note that no such declaration is implied in
2386 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2387 * declared types of functions from the "Conditions"
2388 chapter. (reported by Paul Dietz)
2389 * bug fix: CERROR accepts a function as its first argument.
2390 * bug fix: NTH an NTHCDR accept a bignum as index
2391 arguments. (reported by Adam Warner)
2392 * optimization: character compare routines now optimize comparing
2393 against a constant character. (reported by Gilbert Baumann)
2394 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2395 argument now works properly on the MIPS platform.
2396 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2397 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2398 declared to have a functional type, does not cause an error
2400 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2401 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2402 * fixed bug: if last continuation of a deleted block has a
2403 destination, this destination should be deleted too. (reported by
2405 * fixed a bug in the bootstrap process: the host compiler's values
2406 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2407 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2408 test case from Patrik Nordebo)
2409 * improved the ability of the disassembler on the PPC platform to
2410 provide helpful disassembly notes.
2411 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2412 the class in more cases than previously.
2413 * bug fix: FILE-POSITION now understands :START and :END for
2414 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2415 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2416 8) data. (thanks to David Lichteblau)
2417 * bug fix: it is possible to add a method to a generic function
2418 without lambda list.
2419 * bug fix: reader failed to signal END-OF-FILE inside an
2420 object representation. (reported by Nikodemus Siivola)
2421 * fixed some bugs revealed by Paul Dietz' test suite:
2422 ** LAST and [N]BUTLAST should accept a bignum.
2423 ** condition slot accessors are methods.
2424 ** (VECTOR NIL) is a subtype of STRING.
2426 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2427 * minor incompatible change: some nonsensical specialized lambda
2428 lists (used in DEFMETHOD) which were previously ignored now signal
2430 * minor incompatible change: the system is now aware of the types of
2431 variables in the COMMON-LISP package, and will signal errors for
2432 most violations of these type constraints (where previously they
2433 were silently accepted).
2434 * minor incompatible change: COMPILE-FILE now uses the freedom
2435 afforded (ANSI 3.2.2.3) to use derived function types for
2436 functions defined in the same file. This also permits the system
2437 to warn on static type mismatches and function
2438 redefinition. (Currently it does not work with high DEBUG level.)
2439 * minor incompatible change: VALUES declaration is disabled.
2440 * When issuing notes, the compiler now signals a condition of type
2441 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2442 restart for use in user handlers. It is expected that the
2443 COMPILER-NOTE condition will eventually become a condition
2444 supertype to a hierarchy of note types, which will then be
2445 handleable in a similar fashion. However, at the moment, no such
2446 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2447 but it's an implementation detail, not a classification for the
2449 * Changes in type checking closed the following bugs:
2450 ** type checking of unused values (192b, 194d, 203);
2451 ** template selection based on unsafe type assertions (192c, 236);
2452 ** type checking in branches (194bc).
2453 * A short form of VALUES type specifier has ANSI meaning (it has
2454 increased the number of situations when SBCL cannot perform type
2456 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2457 keywords or constants is permissible.
2458 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2459 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2460 classes. (thanks to Antonio Martinez)
2461 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2462 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2463 argument to a pretty-print stream. Instead, it leaves any such
2464 operations to DESCRIBE-OBJECT methods.
2465 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2467 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2468 lambda lists. (thanks to David Lichteblau)
2469 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2470 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2471 (thanks to Gerd Moellmann)
2472 * fixed bug 246: increased compilation speed of long
2473 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2475 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2477 * DEFCONSTANT now throws a condition of type
2478 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2479 constant to a non-EQL value; CONTINUE and ABORT restarts
2480 respectively change and preserve the value.
2481 * fixed bug 63: The code walker, part of the implementation of CLOS,
2482 is now better at handling symbol macros.
2483 * bug fix: There is no longer an internal implementation type named
2484 CL:LENGTH. (reported by Raymond Toy)
2485 * bug fix: In macro-like defining macros/special operators the
2486 implicit block does not enclose the lambda list.
2487 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2488 enclosed in parentheses) are not suitable as type specifiers, and
2489 their use properly signals an error now.
2490 * bug fix: An argument count mismatch for a type specifier in code
2491 being compiled no longer causes an unhandled error at compile
2492 time, but signals a compile-time warning.
2493 * fixed simple vector readable printing
2494 * bug fix: DESCRIBE takes more care over whether the class
2495 precedence list slot of a class is bound before accessing it.
2496 (reported by Markus Krummenacker)
2497 * bug fix: FORMATTER can successfully compile pretty-printer format
2498 strings which use variants of the ~* directive inside.
2499 * bug fix: SEARCH now applies its TEST predicate to the elements of
2500 the arguments in the correct order. (thanks to Wolfhard Buss)
2501 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2502 in type inference. (thanks to Robert E. Brown)
2503 * bug fix: Reading in symbols with an explicit package name of ""
2504 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2505 * fixed some bugs revealed by Paul Dietz' test suite:
2506 ** NIL is now allowed as a structure slot name.
2507 ** Arbitrary numbers, not just REALs, are allowed in certain
2508 circumstances in LOOP for-as-arithmetic clauses.
2509 ** Multiple class redefinitions before slot access no longer
2510 causes a type error.
2511 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2512 association between the name and a class.
2513 ** Generic functions with non-standard method-combination and over
2514 five methods all of which return constants no longer return NIL
2515 after the first few invocations. (thanks to Gerd Moellmann)
2516 ** CALL-NEXT-METHOD with no arguments now passes the original
2517 values of the arguments, even in the presence of assignment.
2518 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2520 ** DISASSEMBLE works with closures and funcallable instances.
2521 ** ADD-METHOD now returns the generic function, not the new method.
2522 ** FIND-METHOD signals an error if the lengths of the specializers
2523 is incompatible with the generic function, even if the ERRORP
2525 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2526 which its argument is a member.
2527 ** DEFCLASS only redefines the class named by its class-name
2528 argument if that name is the proper name of the class;
2529 otherwise, it creates a new class.
2530 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2531 of the UNBOUND-SLOT condition to the name of the slot.
2532 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2533 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2534 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2535 SLOT-MAKUNBOUND in the specified fashion.
2537 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2538 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2539 cross-compilation host. As a consequence, we can now bootstrap our
2540 way up to SBCL starting with a bare gcc toolchain and human-readable
2541 source code (first the source to CLISP, then the source to SBCL).
2542 * A contributed module containing a partial implementation of the
2543 simple-streams interface has been included. (thanks to Rudi
2545 * A contributed module implementing the RFC1321 Message Digest
2546 Algorithm, known as MD5, has been included.
2547 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2548 no longer has any effect, as the code controlled by this feature
2549 has been deleted. (As far as we know, no-one has ever built using
2550 this feature, and its semantics were confused in any case).
2551 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2552 work on conditions (as required by the ANSI specification),
2553 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2554 expected behaviour on conditions. Users should note, however,
2555 that such behaviour is not required by the ANSI specification,
2556 and so use of this behaviour may render their code unportable.
2557 * fixed some bugs revealed by Paul Dietz' test suite:
2558 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2560 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2561 ** SXHASH on condition objects no longer returns NIL.
2562 ** :ALLOCATION :CLASS slots are better treated; their values are
2563 updated on class redefinition, and initforms inherited from
2564 superclasses are applied.
2565 ** REMOVE-METHOD returns its generic function argument even when
2566 no method was removed.
2567 ** SHARED-INITIALIZE now initializes the values of the requested
2568 slots, including those with :ALLOCATION :CLASS.
2569 ** ALLOCATE-INSTANCE now works on structure classes defined via
2570 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2572 ** SLOT-EXISTS-P now works on conditions, as well as structures
2574 ** MAKE-LOAD-FORM now has the required methods on
2575 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2576 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2577 keyword, and now implements the SLOT-NAMES argument.
2578 ** methods with &OPTIONAL arguments no longer allow too many
2579 arguments to be passed in the call without error.
2580 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2581 option is consistent with the required arguments of the generic
2582 function lambda list.
2583 * bug fix: REQUIRE accepts a string designator. (Thanks to
2585 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2586 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2587 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2588 * bug fix: The compiler now checks for duplicated variables in macro
2590 * bug fix: SETQ on globals returns the correct value.
2591 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2592 not a COMPILER-ERROR (followed by some other strange error on
2593 choosing the CONTINUE restart).
2594 * bug fix: make.sh and friends are now more consistent in the way that
2595 they look for GNU "make".
2597 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2598 * experimental native threads support (on x86 Linux >=2.4 only).
2599 This is not compiled in by default: you need to add :SB-THREAD to
2600 the target features. See the "Beyond ANSI" chapter of the manual
2602 * fix for longstanding nonANSIism: The old distinction between
2603 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2604 The return value from CL:FIND-CLASS is now a CLOS class, and
2605 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2606 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2607 * An interface to the MetaObject Protocol, as described in Kiczales,
2608 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2609 MIT Press, 1991, is available from the SB-MOP package.
2610 * incompatible change: the SB-PCL package should now be considered
2611 a private implementation detail, and no longer a semi-private MOP
2613 * minor incompatible change: due to rearrangement for threads, the
2614 control stack and binding stack are now allocated at arbitrary
2615 addresses instead of being hardcoded per-port. Users affected by
2616 this probably have to be doing advanced things with shared
2617 libraries, and will know who they are.
2618 * minor incompatible change: Previously, all --eval forms used were
2619 processed with READ before any of them were processed with EVAL.
2620 Now each --eval form is processed with both READ and EVAL before
2621 the next --eval form is processed. (Thus package operations like
2622 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2623 work as the user might reasonably expect.)
2624 * minor incompatible change: *STANDARD-INPUT* is now only an
2625 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2627 * minor incompatible change: Y-OR-N-P is now character-oriented,
2628 not line oriented. Also, YES-OR-NO-P now works without errors.
2629 (thanks to Antonio Martinez)
2630 * sb-aclrepl module improvements: an integrated inspector, added
2631 repl features, and a bug fix to :trace command.
2632 * Known functions, which cannot be open coded by the backend, are
2633 considered to be able to check types of their arguments. (fixing
2634 a bug report by Nathan J. Froyd)
2635 * fixed a bug in computing method discriminating functions: It is
2636 now possible to define methods specialized on classes which have
2637 forward-referenced superclasses. (thanks to Gerd Moellmann)
2638 * fixed evaluation order in optional entries (reported by Gilbert
2640 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2641 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2642 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2643 required-by-AMOP NAME argument, as well as CLASS and
2644 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2645 * fixed bug 20: DEFMETHOD can define methods using names that are
2646 not the proper names of classes to designate class specializers.
2647 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2648 Unix file descriptors, instead of blowing up. (thanks to Antonio
2650 * Garbage collection refactoring: user-visible change is that a
2651 call to the GC function during WITHOUT-GCING will not do garbage
2652 collection until the end of the WITHOUT-GCING. If you were doing
2653 this you were probably losing anyway.
2654 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2655 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2656 TYPEP the latter but not the former.
2657 * The compiler issues a full WARNING for calls to undefined functions
2658 with names from the CL package.
2659 * MAP-INTO for a vector destination is open coded. (reported by
2660 Brian Downing on c.l.l)
2661 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2662 documentation string.
2663 * fixed some bugs revealed by Paul Dietz' test suite:
2664 ** COPY-ALIST now signals an error if its argument is a dotted
2666 ** Condition slots are now accessed more correctly in the presence
2667 of multiple initargs for a given slot.
2668 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2669 exclude restarts of the same name associated with a different
2671 ** DEFCLASS of forward-referenced classes with another
2672 forward-referenced class in the superclasses list no longer
2674 ** Condition slots are now initialized once each, not multiple
2675 times. (thanks to Gerd Moellmann)
2676 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2678 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2679 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2680 arguments contain duplicated elements.
2681 ** RESTART-CASE understands local macros.
2682 ** RESTART-CASE associates exactly its own restarts with a condition.
2683 ** ENDP in safe mode checks its argument to be of type LIST.
2684 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2685 in question is unbound.
2686 ** Optimized MAKE-INSTANCE functions no longer cause internal
2687 assertion failures in the presence of duplicate initargs.
2688 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2689 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2690 propagated even in the ctor optimized implementation of
2692 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2694 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2695 * a better implementation of SXHASH on (simple) bit vectors,
2696 measured both in execution speed and in distribution of results
2697 over the positive fixnums, has been installed. Likewise, a better
2698 implementation of EQUAL for simple bit vectors is now available.
2699 * fixed CEILING optimization for a divisor of form 2^k.
2700 * fixed bug 240 (emitting extra style warnings "using the lexical
2701 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2702 by Antonio Martinez)
2703 * fixed SXHASH, giving different results for NIL depending on type
2704 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2705 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2706 variable. (found by Rolf Wester)
2707 * SBCL does not ignore type declarations for special
2708 variables. (reported by rif on c.l.l 2003-03-05)
2709 * some bug fixes in contrib/sb-aclrepl/
2710 * fixed some bugs revealed by Paul Dietz' test suite:
2711 ** a bug in the CONS type specifier, whereby the CAR and CDR
2712 types got intertwined, has been fixed;
2713 ** the type system is now able to reason about the interaction
2714 between INTEGER and RATIO types more completely;
2715 ** APPEND, [N]REVERSE and NRECONC check that those their
2716 arguments, which must be proper lists, are really so;
2717 ** An array specialized to be unable to hold elements has been
2718 implemented, as required -- yes, really -- by ANSI;
2719 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2720 on malformed property lists;
2722 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2723 * incompatible packaging change: in line with Unix convention,
2724 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2725 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2726 * REQUIRE and PROVIDE are now optionally capable of doing something
2727 useful. See the documentation string for REQUIRE.
2728 * infrastructure for a managed SBCL contrib system: contributed
2729 modules in this release include:
2730 ** the ASDF system definition facility;
2731 ** an interface to the BSD Sockets API;
2732 ** an ACL-like convenience interface to the repl;
2733 (thanks to Kevin Rosenberg)
2734 ** an implementation of ROTATE-BYTE, with efficient implementation
2736 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2737 gives a closer approximation to the right answer than previously.
2738 (thanks to Raymond Toy)
2739 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2740 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2741 optional environment arguments, as required by ANSI.
2742 * fixed bugs in other functions taking environment objects, allowing
2743 calls with an explicit NIL environment argument to be compiled
2745 * fixed bug 228: primary return values from
2746 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2747 COMPILE or FUNCTION.
2748 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2749 :TYPE will no longer signal errors on innocuous objects.
2750 * fixed bug 231b: SETQ is better at respecting type declarations in
2751 the lexical environment.
2752 * fixed a bug in DEFCLASS: classes named by symbols with no or
2753 unprintable packages can now be defined.
2754 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2755 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2756 * fixed compiler failure related to checking types of functions.
2757 (reported by Robert E. Brown)
2758 * the compiler is now much more consistent in its error-checking
2759 treatment of bounding index arguments to sequence functions: in
2760 (SAFETY 3) code, errors will be signalled in almost all cases if
2761 invalid sequence bounding indices are passed to functions defined
2762 by ANSI to operate on sequences.
2763 * fixed a bug in the build procedure: documentation of SBCL-specific
2764 packages is now preserved and available in the final Lisp image.
2765 * lifted FDEFINITION lookup out of loops in the implementation of
2766 many list operations. (thanks to Robert E. Brown)
2767 * fixed a bug in the reader: the #n# reader macro now works for
2768 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2769 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2770 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2771 E. Brown and Tony Martinez respectively)
2772 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2773 SLOT-BOUNDP: in particular, it is now able to optimize them much
2774 better, and is now not vulnerable to having packages renamed.
2775 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2776 always called when a slot is not present in an instance. (thanks
2778 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2779 longer signal an error on unreading a character following EOF on
2780 the previous constituent stream. (thanks to Tony Martinez)
2781 * fixed some bugs revealed by Paul Dietz' test suite:
2782 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2783 not just nonnegative fixnums;
2784 ** the logical bit-array operators such as BIT-AND now accept an
2785 explicit NIL for their "opt-arg" argument (to indicate a
2786 freshly-consed result bit-array);
2787 ** ELT now signals an error on an invalid sequence index in safe
2789 ** the type system is now cleverer about negations of numeric
2790 types, and consequently understands the BIGNUM and RATIO types
2792 ** the type system is now cleverer about the interaction between
2793 INTEGER and RATIO types: while bugs still remain, many more
2794 cases are accurately computed;
2795 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2796 if it is in the last clause;
2797 ** CONSTANTLY now correctly returns a side-effect-free function in
2799 ** DECLARE is no longer treated as a special-operator; in
2800 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2801 * incremented fasl file version number due to the change in the
2802 DEFSTRUCT-SLOT-DESCRIPTION structure.
2804 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2805 * minor incompatible change: code processed by the "interpreter" or
2806 EVAL now has a compilation optimization policy of (DEBUG 2)
2807 (changed from (DEBUG 1)) to improve debuggability of interactive
2808 development, and to allow the use of the debug RETURN command in
2810 * an experimental implementation of the RETURN command for the
2811 debugger has been included. (thanks to Frederik Kuivinen)
2812 * fixed bug 62: constraints were not propagated into a loop.
2813 * fixed bug in embedded calls of SORT (reported and investigated by
2815 * fixed some bugs revealed by Paul Dietz' test suite:
2816 ** printing and reading of arrays with some dimensions having
2817 length 0 (thanks to Gerd Moellmann);
2818 ** BOA constructor with &AUX argument without a default value does
2819 not cause a type error;
2820 ** CONSTANTP now returns true for all self-evaluating objects.
2822 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2823 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2824 accessors that are related by inheritance, as specified in the
2825 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2826 Valtteri Vuorikoski)
2827 * The compiler is now able to inline functions that were defined in
2828 a complex lexical environment (e.g. inside a MACROLET).
2829 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2831 * Support for the upcoming FreeBSD-5.0 release has been included.
2832 (thanks to Dag-Erling Smorgrav)
2833 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2834 effect when it is not in a toplevel context.
2835 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2836 stand a better chance of being correct. (thanks to Gerd
2838 * fixed bug in COERCE, which now signals an error on coercing a
2839 rational to a bounded real type which excludes the expected
2841 * The compiler is now able to derive types more accurately from the
2842 COERCE and COMPILE functions.
2843 * fixed bug 223: functional binding is considered to be constant
2844 only for symbols in the CL package.
2845 * fixed bug 231: SETQ did not check the type of a variable being set
2846 (reported by Robert E. Brown)
2847 * A new optimization for MAKE-INSTANCE has been included, fixing
2848 various bugs (including relating to :ALLOCATION :CLASS slots and
2849 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2851 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2852 ** As required by ANSI, LOOP now disallows anonymous collection
2853 clauses such as COLLECT I in conjunction with aggregate boolean
2854 clauses such as THEREIS (= I 1);
2855 ** LOOP now signals an error when any variable is reused in the
2856 same loop (including the potentially useful construct analogous
2857 to WITH A = 1 WITH A = (1+ A);
2858 ** IT is only a special loop symbol within the first clause of a
2859 conditional loop clause;
2860 ** LOOP with a typed iteration variable over a hashtable now
2861 signals a type error iff it should.
2862 * fixed some other bugs revealed by Paul Dietz' test suite:
2863 ** FILE-STREAM now names the class previously known as FD-STREAM;
2864 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2865 argument) no longer signals an error;
2866 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2867 of the slot symbol, rather than using the current package
2868 ((:CONC-NAME "") continues to intern the slot's name in the
2870 * incremented fasl file version number, because of the incompatible
2871 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2872 of the new implementation of DEFINE-COMPILER-MACRO.
2874 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2875 * Support for building SBCL for MIPS platforms running in
2876 little-endian mode has now been checked in, and basic
2877 functionality on said platforms verified.
2878 * minor incompatible change: PCL now records the pathname of a file
2879 in which methods and the like are defined, rather than its
2881 * minor incompatible change: TRUENAME now considers the truename of
2882 a file naming a directory to be the pathname with :DIRECTORY
2883 component indicating that directory.
2884 * minor incompatible change: a NAMED clause in the extended form of
2885 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2886 reason for the previous behaviour is unclear.
2887 * more systematization and improvement of CLOS and MOP conformance
2888 in PCL (thanks to Gerd Moellman and Pierre Mai):
2889 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2891 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2892 primary methods with no specializers;
2893 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2895 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2896 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2897 FORWARD-REFERENCED-CLASSes; error reporting on
2898 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2900 ** SXHASH on CLOS instances now uses a slot internal to the
2901 instance to return different numbers on distinct instances,
2902 while preserving the same return value through invocations of
2904 ** DEFMETHOD signals errors when methods with longer incongruent
2905 lambda lists are added to generic functions;
2906 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2907 CLASS, as specified in AMOP;
2908 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2909 on the order returned by the primary method for classes of
2910 class STANDARD-CLASS;
2911 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2912 * fixed some bugs shown by Paul Dietz' test suite:
2913 ** DOLIST puts its body in TAGBODY;
2914 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2916 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2917 value producing form;
2918 ** if more variables are given to PROGV than values, extra
2919 variables are bound and made to have no value;
2920 ** NSUBSTITUTE on list arguments gets the right answer with
2922 ** ELT signals an error of type TYPE-ERROR when the index argument
2923 is not a valid sequence index;
2924 ** LOOP signals (at macroexpansion time) an error of type
2925 PROGRAM-ERROR when duplicate variable names are found;
2926 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2927 ** FUNCALL of special-operators now cause an error of type
2929 ** PSETQ now works as required in the presence of side-effecting
2930 symbol-macro places;
2931 ** NCONC accepts any object as its last argument;
2932 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2934 ** loop-for-as-package does not require a package to be explicitely
2936 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2938 * fixed bug 166: compiler preserves "there is a way to go"
2939 invariant when deleting code.
2940 * fixed bug 172: macro lambda lists with required arguments after
2941 &REST arguments now cause an error to be signalled. (thanks to
2943 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2944 bug. (thanks to Matthew Danish)
2945 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2947 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2948 arguments are lexically rebound. (thanks to Gerd Moellmann and
2950 * fixed bug 194: error messages are now more informative when there
2951 is no primary method applicable in a call to a generic
2952 function. (thanks to Gerd Moellmann)
2953 * fixed bug in command line argument checking (thanks to Julian
2955 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2956 (thanks to Matthew Danish)
2957 * incremented fasl file version number, because of the
2958 SXHASH-related changes in the layout of CLOS data structures
2960 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2961 * minor incompatible change: The runtime (the Unix executable named
2962 "sbcl") is now much pickier about the .core files it will load.
2963 Essentially it now requires .core files to descend from the same
2964 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2965 as the runtime does. (The intent is to prevent the crashes which
2966 can occur, and which can even be reported as mysterious failures,
2967 when people patch the sources or change the build parameters
2968 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2969 sbcl and .core files.)
2970 * fixed bug: VALUES-LIST is no longer optimized away.
2971 * fixed bug 142: The FFI conversion of C string values to Lisp
2972 string values no longer conses excessively. (thanks to Nathan
2973 Froyd porting Raymond Toy's fix to CMU CL)
2974 * began to systematize and improve MOP conformance in PCL (thanks to
2975 Nathan Froyd, Gerd Moellman and Pierre Mai):
2976 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2978 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2979 ** FINALIZE-INHERITANCE is now called on class finalization;
2980 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2981 argument precedence order.
2982 * fixed bug 202: The compiler no longer fails on functions whose
2983 derived types contradict their declared type.
2984 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2985 so it can be non-toplevel.
2986 * The fasl file version number has changed (because of the new
2987 implementation of DEFMACRO).
2988 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2989 safe code, that any length requirement by their type-specifier
2990 argument is valid. The exceptions to this are described in bug
2992 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2993 functions now check the types of their inputs as required by ANSI.
2994 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2995 introduced symbol is DECLAREd to be SPECIAL.
2996 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2997 * fixed bug: :COUNT argument to sequence functions may be negative
2998 * fixed bug: body of DO-SYMBOLS may contain declarations
2999 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3000 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3001 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3002 symbol macro only once
3003 * fixed printing of call frame when argument list is unavailable
3004 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3005 * compiler no longer signals WARNING on unknown keyword
3008 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3009 * A beta-quality port to the mips architecture running Linux,
3010 based on the old CMUCL backend, has been made. It has been tested
3011 on a big-endian kernel, and works sufficiently well to be able to
3012 rebuild itself; it has not been tested in little-endian mode.
3013 * fixed an inconsistency between gencgc.c and purify.c which made
3014 dumping/loading .core files unreliable
3015 * fixed bug 120a: The compiler now deals correctly with IFs where
3016 the consequent is the same as the alternative, instead of
3017 misderiving the return type. (thanks to Alexey Dejneka)
3018 * fixed bug 113: Logical pathnames are now dumpable (the logical
3019 host is resolved at load-time, throwing an error if it is not
3021 * fixed bug 174: FORMAT's error message is slightly clearer when a
3022 non-printing character is used in a format directive.
3023 * fixed several bugs in compiler checking of type declarations, i.e.
3024 violations of the Python "declarations are assertions" principle
3025 (thanks to Alexey Dejneka)
3026 * fixed several bugs in PCL's error checking (thanks to Gerd
3028 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3030 * fixed bug in compilation of functions as first class values
3031 (thanks to Antonio Martinez-Shotton)
3032 * The compiler's handling TYPE-ERRORs which it can prove will
3033 inevitably happen at runtime has been cleaned up and corrected
3034 in several ways. (thanks to Alexey Dejneka)
3035 * improved argument type checking for various basic arithmetic
3036 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3037 lavished on them in the past that they can be compiled in many
3038 ways in different special cases
3039 * fixed bug 181: compiler checks validity of user supplied type
3041 * cleaned up code flushing in optimization: Function calls which
3042 should signal errors for safety purposes (e.g. which ANSI says
3043 should signal errors when their arguments are of incorrect type)
3044 are no longer optimized away.
3045 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3046 * incremented fasl file version number, because changes in the
3047 implementation of sequence functions like COERCE caused
3048 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3049 in old inline expansions) to become undefined. (Actually these
3050 changes were later undone, so we might very well be binary
3051 compatible with 0.7.7 after all, but leaving the version number
3052 incremented seemed like the simplest and most conservative
3055 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3056 * An alpha-quality port to the parisc architecture running Linux,
3057 based on the old CMUCL backend, has been made. This, even more so
3058 than the other backends, should be considered still a work in
3059 progress; known problems include that the Linux kernel in 64-bit
3060 mode does not propagate the correct sigcontext structure to
3061 userspace, and consequently SBCL on a parisc64 kernel will not
3063 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3064 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3065 suppressed.) Also now that the compiler is looking at declarations
3066 in the environment, it checks optimization declarations as well,
3067 and suppresses inlining when (> DEBUG SPEED).
3068 * More fixes have been made to treatment of floating point exception
3069 treatment and other Unix signals. In particular, floating point
3070 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3071 * The detection and handling of control stack exhaustion (infinite
3072 or very deeply nested recursion) has changed. Stack exhaustion
3073 detection is now done by write-protecting pages at the OS level
3074 and applies at all optimization settings; when found, a
3075 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3076 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3077 be caught using IGNORE-ERRORS.
3078 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3079 correctly when transforming recursive LABELS functions to LETs.
3080 (thanks to Alexey Dejneka)
3081 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3082 that are names of constants or global variables.
3083 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3084 alien routines with docstrings.
3085 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3086 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3088 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3089 (thanks to Eric Marsden)
3090 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3091 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3092 * Bug fix: The SPARC backend can now compile functions involving
3093 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3094 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3095 to a functions expecting a PATHNAME-DESIGNATOR.
3096 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3097 lambda lists. (thanks to Alexey Dejneka)
3098 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3099 functions so that EXPORTing the name of the class doesn't cause
3100 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3101 to Antonio Martinez for reporting this)
3102 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3103 constant-folded without failing an assertion. (thanks to Einar
3104 Floystad Dorum for reporting this)
3105 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3106 types are now dealt with more correctly by the compiler.
3107 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3108 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3109 behaviour analogous to RENAME-FILE. This puts its behaviour more
3110 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3112 * The fasl file version number has changed again. (because of the
3113 bug fix involving the names of PCL MAKE-INSTANCE functions)
3115 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3116 * bug fix: Floating point exceptions are treated much more
3117 consistently on the x86/Linux and PPC/Linux platforms.
3118 * Array initialization with :INITIAL-ELEMENT is now much faster for
3119 cases when the compiler cannot open code the array creation, but
3120 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3121 array accesses have also seen a speed increase.
3122 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3123 specified. (This was at the root of some bad interactions between
3124 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3126 * bug fix: Internal error arguments for undefined functions are now
3127 computed correctly on the PPC/Linux platform.
3128 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3129 Raymond Toy's patch for CMU CL)
3130 * Support for the Solaris 9 operating environment has been included
3131 (thanks to Daniel Merritt)
3132 * A very ugly but hopefully complete draft of the missing FFI chapter
3133 of the manual has been created by reformatting the corresponding
3134 CMU CL manual chapter into (currently very ugly and incoherent)
3135 DocBook and bringing it up to date for SBCL behavior. Thus, the
3136 manual is now essentially complete, at least by my extreme
3137 once-and-only-once standards, whereby it's acceptable to refer to
3138 the doc strings of SB-EXT functions as the primary documentation.
3139 * The fasl file version number has changed again, due to cleanup of
3140 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3141 is no longer a static symbol.)
3143 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3144 * SBCL now builds with OpenMCL (version 0.12) as the
3145 cross-compilation host; also, more progress has been made toward
3146 bootstrapping under CLISP.
3147 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3149 * bug 158 fixed: The compiler can now deal with integer loop
3150 increments different from 1; fixing this turned out also to fix
3152 * bug 169 fixed: no more bogus warnings about using lexical bindings
3153 despite the presence of perfectly good SPECIAL declarations (thanks
3154 to David Lichteblau)
3155 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3156 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3157 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3159 * bug 180 fixed: Method combination specifications no longer ignore
3160 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3161 * bug fix: Structure type predicate functions now check their argument
3162 count as they should.
3163 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3164 correctly. (thanks to Pierre Mai)
3165 * minor incompatible change: The --noprogrammer option is deprecated
3166 in favor of the new --disable-debugger option, which is very similar.
3167 (The major difference is that it takes effect at a slightly different
3168 time at startup, causing handling of errors in --sysinit and
3169 --userinit files will be affected differently.) The
3170 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3171 been added to allow this functionality to be controlled from ordinary
3172 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3173 maintainers, who might want to run non-interactive scripts to
3174 build SBCL cores which will later be used interactively.)
3175 * minor incompatible change: The LOAD function no longer, when given
3176 a wild pathname to load, loads all files matching that pathname.
3177 Instead, an error of type FILE-ERROR is signalled.
3179 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3180 * bug 147 fixed: The compiler preserves its block link/count
3181 invariants more correctly now so that it doesn't crash. (thanks
3183 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3185 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3186 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3187 no-value-for-this-character as NIL (as specified by ANSI).
3188 * HOST-NAMESTRING on physical pathnames now returns a string that is
3189 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3190 (thanks to Christophe Rhodes)
3191 * The Alpha port handles icache flushing more correctly. (thanks to
3193 * More progress has been made toward bootstrapping under CLISP. (thanks
3194 to Christophe Rhodes)
3195 * The fasl file format has changed again, because dynamic loading
3196 on OpenBSD (which has non-ELF object files) motivated some cleanups
3197 in the way that foreign symbols are transformed and passed around.
3198 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3199 is no longer treated as whitespace by the reader, but instead as
3200 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3201 |A
\7fB|, instead of A as it used to.
3203 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3204 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3205 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3206 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3207 this as of 0.7.1.45, but was left out of the previous news section
3208 (thanks to Dan Barlow)
3209 * SBCL now runs on the Solaris operating system on SPARC architectures
3210 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3211 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3212 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3213 * SPARC backend cleanups, allowing builds of cores optimized for V8
3214 and V9 SPARCS, and also emission of code targeted to a particular
3215 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3217 * SBCL is closer to bootstrapping under CLISP, thanks to various
3218 fixes by Christophe Rhodes.
3219 * The fasl file format has changed again, to allow the compiler's
3220 INFO database to support symbol macros.
3221 * The user manual (in doc/) is formatted into HTML more nicely.
3222 (thanks to coreythomas)
3223 * The system is smarter about SUBTYPEP relationships, especially
3224 those involving NOT types (including types such as ATOM which are
3225 represented internally using NOT types). Thus SUBTYPEP is less
3226 likely to return (VALUES NIL NIL) in general, and in particular
3227 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3229 * The fasl file format has changed again, because the internal
3230 representation of types now includes a new slot to support the new
3231 SUBTYPEP-of-NOT-types logic.
3232 * (not a change in the main branch of SBCL, but a related prototype
3233 which can hopefully be merged into the main branch of SBCL in the
3234 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3235 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3236 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3237 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3238 dispatch macro character combinations. (thanks to Alexey Dejneka)
3239 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3240 that superficially look like logical namestrings correctly.
3241 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3243 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3244 default supplied-p) for &optional and &key arguments. (thanks to
3247 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3248 * incompatible change: The compiler is now less aggressive about
3249 tail call optimization, doing it only when (> SPACE DEBUG) or
3250 (> SPEED DEBUG). (This is an incompatible change because there are
3251 programs which relied on the old CMU-CL-style behavior to optimize
3252 away their unbounded recursion which will now die of stack overflow.)
3253 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3254 for non-GENCGC systems has been increased to 20M (since that
3255 seems much closer to the likely performance optimum for modern
3256 systems than the old 4M value was)
3257 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3258 and *DEBUG-PRINT-LEVEL*
3259 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3260 of CMU CL's support for SPARC, and various endianness and other
3261 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3262 * new syntactic sugar for the Unix command line: --load foo.bar is now
3263 an alternate notation for --eval '(load "foo.bar")'.
3265 ** The system now detects stack overflow and handles it gracefully,
3266 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3267 optimization settings. (This is a good thing in general, and
3268 its introduction in this version should be particularly timely
3269 for anyone whose code fails because of suppression of tail
3271 ** The system now hunts for the C variable "environ" in a more
3272 devious way, to avoid segfaults when the C library version differs
3273 between compile time and run time. (thanks to Christophe Rhodes)
3274 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3275 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3276 ** The compiler no longer issues bogus style warnings for undefined
3277 classes in the same source file as the DEFCLASSes which defined
3278 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3280 ** fixes in CONDITION class precedence list for undefined function
3281 errors (thanks to Alexei Dejneka)
3282 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3283 correctly. (thanks to Dan Barlow)
3284 ** portability fixes aiming at bootstrapping under CLISP (thanks
3285 to Dave McDonald and Christophe Rhodes)
3286 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3287 ** fixes in type translation and and type inference (thanks to
3289 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3290 ** various fixes in the new ports (thanks to Dan Barlow)
3291 * several changes related to debugging:
3292 ** suppression of tail recursion, as noted above
3293 ** stack overflow detection, as noted above
3294 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3295 is now the default. (For some time encapsulation has been more
3296 reliable than the breakpoint-based :ENCAPSULATE NIL
3297 implementation, at least on X86 systems; and I just noticed that
3298 encapsulation also seems closer to the spirit of the ANSI
3301 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3303 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3304 up properly again. (There was a packaging bug in 0.7.0 which
3305 left their definitions in SB-SYS::LOAD-FOREIGN and
3306 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3307 vital for most things which interface to C-level interfaces,
3308 like extensions working with sockets or databases or
3309 Perl-compatible regexes or whatever, and the need to fix
3310 this bug is the main reason that 0.7.1 was released so
3312 ** DEFGENERIC is now choosier about the methods it redefines, so that
3313 reLOADing a previously-LOADed file containing DEFGENERICs does
3314 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3315 cycle now works as it should. (thanks to Alexey Dejneka)
3316 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3317 fixed by Christophe Rhodes. (He actually submitted this patch
3318 months ago, and I delayed until after 0.7.0.)
3319 ** Bug 111 (internal compiler confusion about runtime checks on
3320 FUNCTION types) was fixed by Alexey Dejneka.
3321 * Some internal cleanups (getting rid of variables which aren't
3322 needed now that the byte interpreter is gone) caused the fasl
3323 file format number to change again.
3325 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3326 * major incompatible change: The default fasl file extension, i.e. the
3327 default extension for files produced by COMPILE-FILE, has changed
3328 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3330 ** There are many changes in the implementation of the compiler.
3331 SBCL is now essentially a compiler-only implementation of ANSI
3332 Common Lisp. EVAL still "interprets" a few special cases, but
3333 almost all the interesting cases are handled by creating
3334 a LAMBDA expression, calling COMPILE on it, then calling
3335 FUNCALL on the result.
3336 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3337 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3338 Since the code is newer, there might still be some new bugs
3339 (though not as many as before Martin Atzmueller's fixes:-). But
3340 the new code is substantially simpler and clearer, and hopefully
3341 any remaining bugs will be simpler, less fundamental, and more
3342 fixable then the bugs in the old code.
3343 ** The revised compiler is still a little unsteady on its feet.
3345 *** The debugging information it produces (particularly the names
3346 of FUNCTION objects) is sometimes much less useful than what
3347 the old compiler produced.
3348 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3349 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3350 within a MACROLET) has been temporarily weakened.
3351 ** There are new compiler optimizations for various functions:
3352 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3353 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3354 *** the math functions TRUNCATE, FLOOR, and CEILING
3355 *** the function-of-all-trades COERCE
3356 Mostly these should be transparent, but there's one
3357 potentially-annoying problem (bug 117): when the compiler
3358 inline-expands a function and does type analysis on the result,
3359 it can create control paths which have type mismatches, and
3360 when it can't prove that those control paths aren't taken,
3361 it will issue WARNINGs about the type mismatches. This is
3362 a particular problem in practice for the new sequence functions.
3363 It's not clear how this should be fixed, and for now, a
3364 workaround is given in the entry for 117 in the BUGS file.
3365 ** (Because of the interaction between the two previous items --
3366 occasional inlining problems and new inline expansions -- some
3367 of the new sequence function optimizations won't really kick in
3368 completely until debugging information, and then inlining, are
3369 straightened out in some future version.)
3370 * minor incompatible changes:
3371 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3372 DIRECTORY no longer implicitly promotes NIL slots of its
3373 pathname argument to :WILD. In particular, when you ask for the
3374 contents of a directory (which you used to be able to do without
3375 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3376 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3377 ** changes in behavior that ANSI explicitly defines to be
3378 implementation dependent:
3379 *** The new compiler-only implementation still conforms with ANSI,
3380 but acts a little different than before. Besides the obvious
3381 changes in performance tradeoffs (that the cost per form passed
3382 to EVAL has gone up, and the cost per form executed by EVAL
3383 has gone down), the behavior of the system changes a little
3384 because there are no longer any interpreted function objects.
3385 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3386 e.g. doing COMPILE on the output of interactive DEFUN is
3388 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3390 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3391 has changed from (:CL) to NIL.
3392 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3393 e.g. #\Newline and #\Tab, don't have names specified in the
3394 ANSI Common Lisp standard, is now based on their ASCII symbolic
3395 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3396 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3397 are no longer used for output.
3398 ** changes in internal implementation constants:
3399 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3400 4 million. (If your application spends a lot of time GCing and
3401 you have a lot of RAM, you might want to experiment with
3402 increasing it even more.)
3403 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3404 However, almost all old code should still continue to work without
3405 immediate update, as SB-C-CALL is now a (deprecated) nickname
3407 ** Old operator names in the style DEF-FOO are now deprecated in
3408 favor of new corresponding names DEFINE-FOO, for consistency with
3409 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3410 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3411 internal symbols, but a few supported extensions like
3412 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3413 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3414 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3415 etc. as you get deeper into recursive calls to the debugger
3416 command loop, instead of the old "5]", "5]]", "5]]]"
3417 sequence. (I was motivated to do this when squabbles between
3418 ILISP and SBCL left me very deeply nested in the debugger. In the
3419 short term, this change will probably provoke more ILISP/SBCL
3420 squabbles, but hopefully it will be an improvement in the long run.)
3421 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3422 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3423 function, since declaiming return types involving aliens is
3424 (1) annoyingly messy to do by hand and (2) vital to efficient
3425 compilation of code which calls such functions.
3426 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3427 longer reexported by the SB-EXT package. They're solely useful
3428 for alien code, so it seems more logical that you should get
3429 them from the SB-ALIEN package, not in SB-EXT.
3430 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3431 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3432 features. Instead, the code that they used to control is always
3433 built into the system.
3434 * many other bug fixes
3435 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3436 advantage of the new EVAL-WHEN stuff and to clean them up in
3437 general, and they are now more ANSI-compliant in a number of
3438 ways. Martin Atzmueller is responsible for a lot of this.
3439 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3441 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3442 *** correct ERROR type for various file operations
3443 *** some fixes for Lisp streams
3444 *** DEFMETHOD syntax checking
3445 *** changing old weird representation of debug information as
3446 strings (which, among their other deficiencies, don't transform
3447 correctly when you rename packages, and don't change their
3448 print representation when you change things like *PACKAGE*
3449 and *PRINT-LENGTH*) to symbols and lists of symbols
3450 He also made several improvements and fixed several bugs in DESCRIBE.
3451 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3453 *** misbehavior of WRITE-STRING/WRITE-LINE
3454 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3455 and several other LOOP problems as well
3456 *** DIRECTORY when similar filenames are present
3457 *** DEFGENERIC with :METHOD options
3458 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3459 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3460 *** argument ordering in FIND with :TEST option
3461 *** mishandled package designator argument in APROPOS-LIST
3462 *** various problems in the backquote readmacro
3463 *** a bug in APROPOS
3464 *** probably some others that I'm not describing very well here,
3465 since the CVS log documents them by reference to sbcl-devel
3466 messages, and the SourceForge archives aren't working well.:-(
3467 ** Dan Barlow improved the Alpha port (and is making progress on the
3468 PPC port, for those of you who think different).
3469 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3470 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3471 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3472 continues to work on the SPARC port (for those of you in a position
3473 to look down upon our little PC-compatible boxes from a great height).
3474 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3475 and uses that copy, rather than the current dynamic value, when
3476 it's trying to decide whether to truncate output. Thus e.g.
3477 (let ((*print-lines* 50))
3478 (pprint-logical-block (stream nil)
3480 (let ((*print-lines* 8))
3481 (print (aref possiblybigthings i) stream)))))
3482 should now truncate the logical block only at 50 lines, instead of
3483 often truncating it at 8 lines, as it did before.
3484 * The doc/cmucl/ directory, containing old CMU CL documentation
3485 from the time of the fork, is no longer part of the base system.
3486 SourceForge has shut down its anonymous FTP service, and with it
3487 my original plan for distributing the old CMU CL documentation
3488 there. For now, if you need these files you can download an old
3489 SBCL source release and extract them from it.
3490 * The fasl file version number changed again, for dozens of reasons,
3491 some of which are apparent above.
3493 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3494 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3495 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3496 patch, so that DISASSEMBLE output is much nicer.
3497 * The code in the SB-PROFILE package now seems reasonably stable.
3498 I still haven't decided what the final interface should look like
3499 (I'd like PROFILE to interact cleanly with TRACE, since both
3500 facilities use function encapsulation) but if you have a need
3501 for profiling now, you can probably use it successfully with
3502 the current CMU-CL-style interface.
3503 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3504 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3505 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3506 link now returns the dangling link itself, and for similar
3507 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3508 link itself. (In these cases the old code signalled an error and
3509 looped endlessly, respectively.) Thus, DIRECTORY now works even
3510 in the presence of dangling and cyclic symbolic links.
3511 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3512 is now a supported extension again, since the consensus on
3513 sbcl-devel was that it can be useful for ordinary development
3514 work, not just for debugging SBCL itself.
3515 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3516 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3517 that currently-defined functions might be redefined later with
3518 different return types.
3519 * Hash tables can be printed readably, as inspired by CMU CL code
3520 of Eric Marsden and SBCL code of Martin Atzmueller.
3521 * better error handling in CLOS method combination, thanks to
3522 Martin Atzmueller porting Pierre Mai's CMU CL patches
3523 * more overflow fixes for >16Mbyte I/O buffers
3524 * A bug in READ has been fixed, so that now a single Ctrl-D
3525 character suffices to cause end-of-file on character streams.
3526 In particular, now you only need one Ctrl-D at the command
3527 line (not two) to exit SBCL.
3528 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3530 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3531 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3532 does the right thing.
3533 * The implementation of some type tests, especially for CONDITION
3534 types, is now tidier and maybe faster, due to CMU CL code
3535 originally by Douglas Crosher, ported by Martin Atzmueller.
3536 * Some math functions have been fixed, and there are new
3537 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3538 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3539 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3540 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3541 likely to be built into sbcl-0.7.0.)
3542 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3543 CL, ported to SBCL by Martin Atzmueller)
3544 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3545 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3546 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3547 * There's a new slam.sh hack to shorten the edit/compile/debug
3548 cycle for low-level changes to SBCL itself, and a new
3549 :SB-AFTER-XC-CORE target feature to control the generation of
3550 the after-xc.core file needed by slam.sh.
3551 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3552 COMPILE-FILE is no longer supported, so that now every function
3553 gets an entry point, so that block compilation looks a little
3554 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3555 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3556 returns the number of bytes consed since the system started,
3557 rather than the number consed since the first time the function
3558 was called. (The new definition parallels ANSI functions like
3559 CL:GET-INTERNAL-RUN-TIME.)
3560 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3561 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3562 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3563 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3564 * Fasl file version numbers are now independent of the target CPU,
3565 since historically most system changes which required version
3566 number changes have affected all CPUs equally. Similarly,
3567 the byte fasl file version is now equal to the ordinary
3570 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3571 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3572 is no longer recognized. I apologize for this, because it was
3573 listed in SB-EXT as a supported extension, but I found that
3574 its existing behavior was poorly specified, as well as incorrectly
3575 specified, and it looked like too much of a mess to straighten it
3576 out. I have enough on my hands trying to get ANSI stuff to work..
3577 * many patches ported from CMU CL by Martin Atzmueller, with
3578 half a dozen bug fixes in pretty-printing and the debugger, and
3579 half a dozen others elsewhere
3580 * fixed bug 13: Floating point infinities are now supported again.
3581 They might still be a little bit flaky, but thanks to bug reports
3582 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3583 as flaky as they were.
3584 * The --noprogrammer command line option is now supported. (Its
3585 behavior is slightly different in detail from what the old man
3586 page claimed it would do, but it's still appropriate under the
3587 same circumstances that the man page talks about.)
3588 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3589 are now supported, and enabled by default. Thus, the compiler can
3590 handle many floating point and complex operations much less
3591 inefficiently. (Thus e.g. you can implement a complex FFT
3593 * The compiler now detects type mismatches between DECLAIM FTYPE
3594 and DEFUN better, and implements CHECK-TYPE more correctly, and
3595 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3596 thanks to patches from Martin Atzmueller.
3597 * various fixes to make the cross-compiler more portable to
3598 ANSI-conforming-but-different cross-compilation hosts (notably
3599 Lispworks for Windows, following bug reports from Arthur Lemmens)
3600 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3601 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3602 fixed thanks to Pierre Mai's CMU CL patches.
3603 * fixes to keep the system from overflowing internal counters when
3604 it tries to use i/o buffers larger than 16M bytes
3605 * fixed bug 45a: Various internal functions required to support
3606 complex special functions have been merged from CMU CL sources.
3607 (When I was first setting up SBCL, I misunderstood a compile-time
3608 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3609 * improved support for type intersection and union, fixing bug 12
3610 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3611 more obscure bugs as well
3612 * some steps toward byte-compiling non-performance-critical
3613 parts of the system, courtesy of patches from Martin Atzmueller
3614 * Christophe Rhodes has made some debian packages of sbcl at
3615 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3616 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3617 stable, but are nonetheless usable. When he's ready, I'd be happy
3618 to add them to the SourceForge "File Releases" section. (And if
3619 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3620 * new fasl file format version number (because of changes in
3621 internal representation of (OR ..) types to accommodate the new
3622 support for (AND ..) types, among other things)
3624 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3625 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3627 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3629 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3630 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3631 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3632 * DESCRIBE now works on structure objects again.
3633 * Most function call argument type mismatches are now handled as
3634 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3635 know whether the function will be redefined before the call is
3636 executed. (The compiler could flag local calls with full WARNINGs,
3637 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3638 it doesn't keep track of enough information to know whether calls
3639 are local in this sense.)
3640 * Compiler output is now more verbose, with messages truncated
3641 later than before. (There should be some supported way for users
3642 to override the default verbosity, but I haven't decided how to
3643 provide it yet, so this behavior is still controlled by the internal
3644 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3645 src/compiler/ir1util.lisp.)
3646 * Fasl file format version numbers have increased again, because
3647 support for the Gray streams extension changes the layout of the
3648 system's STREAM objects.
3649 * The Gray subclassable streams extension now works, thanks to a
3650 patch from Martin Atzmueller.
3651 * The full LOAD-FOREIGN extension (not just the primitive
3652 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3653 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3654 but like most other programs, it defaults to copying the Unix
3655 environment from the original process instead of starting the
3656 new process in an empty environment.
3657 * Extensions which manipulate the Unix environment now support
3658 an :ENVIRONMENT keyword option which doesn't smash case or
3659 do other bad things. The CMU-CL-style :ENV option is retained
3660 for porting convenience.
3661 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3662 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3664 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3666 * A patch from Martin Atzmueller seems to have solved the SIGINT
3667 problem, and as far as we know, signal-handling now works cleanly.
3668 (If you find any new bugs, please report them!)
3669 * The system no longer defaults Lisp source file names to types
3670 ".l", ".cl", or ".lsp", but only to ".lisp".
3671 * The compiler no longer uses special default file extensions for
3672 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3673 seems to expect a single default extension for all compiled code,
3674 and there's no compelling reason to try to stretch the standard
3675 to allow two different extensions.) Instead, byte-compiled files
3676 default to the same extension as native-compiled files.
3677 * Fasl file format version numbers have increased again, because
3678 a rearrangement of internal implementation packages made some
3679 dumped symbols in old fasl files unreadable in new cores.
3680 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3681 many fewer weird special cases.
3682 * Bug #17 (differing COMPILE-FILE behavior between logical and
3683 physical pathnames) has been fixed, and some related misbehavior too,
3684 thanks to a patch from Martin Atzmueller.
3685 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3686 by Tim Moore, ported to SBCL by Martin Atzmueller.
3687 * Martin Atzmueller fixed several filesystem-related problems,
3688 including bug #36, in part by porting CMU CL patches, which were
3689 written in part by Paul Werkowski.
3690 * More compiler warnings in src/runtime/ are gone, thanks to
3691 more patches from Martin Atzmueller.
3692 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3695 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3697 * DESCRIBE now works on CONDITION objects.
3698 * The debugger now handles errors which arise when trying to print
3699 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3701 * The build system now uses an additional file, customize-target-features.lisp,
3702 to allow local modifications to the target *FEATURES* list. (The point of
3703 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3704 debugging features enabled, without having to worry about propagating it
3705 into everyone's system when I do a "cvs update".) When no
3706 customize-target-features.lisp file exists, the target *FEATURES* list
3707 should be constructed the same way as before.
3708 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3709 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3710 warn about behavior which is undefined under the ANSI spec. Note: This
3711 is specified by ANSI, but it's not very popular with programmers.
3712 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3713 macro in the SBCL sources for an example of a workaround which you
3714 might use to make portable ANSI-standard code which does what you want.
3715 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3716 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3717 when it's used as a non-toplevel form. (This is required in order
3718 to implement the DEFCONSTANT-EQX macro.)
3719 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3720 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3721 funny things, due to bugs in the implementation of EVAL-WHEN
3722 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3723 non-toplevel DEFCONSTANTs are uncommon.)
3724 * The core file version number and fasl file version number have been
3725 incremented, because the old noncompliant DEFCONSTANT behavior involved
3726 calling functions which no longer exist, and because I also took the
3727 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3728 * fixed bug 1 (error handling before read-eval-print loop starts), and
3729 redid debugger restarts and related debugger commands somewhat while
3731 ** The QUIT debugger command is gone, since it did something
3732 rather different than the SB-EXT:QUIT command, and since it never
3733 worked properly outside the main toplevel read/eval/print loop.
3734 Invoking the new TOPLEVEL restart provides the same functionality.
3735 ** The GO debugger command is also gone, since you can just invoke
3736 the CONTINUE restart directly instead.
3737 ** The TOP debugger command is also gone, since it's redundant with the
3738 FRAME 0 command, and since it interfered with abbreviations for the
3740 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3742 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3743 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3744 some porting work by Martin Atzmueller)
3745 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3746 with a boatload of patches which he ported to SBCL. Now that those
3748 ** The system tries to make sure that its low-priority messages
3749 are prefixed by semicolons, to help people who like to use
3750 syntax highlighting in their ILISP buffer. (This patch
3751 was originally due to Raymond Toy.)
3752 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3753 patches originally written by Raymond Toy.
3754 ** The compiler understands coercion between single-value and
3755 multiple-VALUES type expressions better, getting rid of some very
3756 weird behavior, thanks to patches originally by Robert MacLachlan
3757 and Douglas Crosher.
3758 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3759 lambda lists, thanks to a patch originally by Pierre Mai.
3760 ** The system no longer bogusly warns about "abbreviated type
3762 ** The compiler gets less confused by inlining and RETURN-FROM,
3763 thanks to some patches originally by Tim Moore.
3764 ** The system no longer hangs when dumping circular lists to fasl
3765 files, thanks to a patch originally from Douglas Crosher.
3766 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3767 undefined function error.
3768 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3769 in structure slot initforms) for now, documented workaround instead:-|
3770 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3771 * fixed bug 5: added stubs for various Gray stream functions called
3772 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3773 installed, at least appropriate type errors are generated
3774 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3775 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3776 consistently in DEFMETHOD forms.
3777 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3778 it doesn't seem to affect SBCL after all
3779 * The C runtime system now builds with better optimization and many
3780 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3782 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3784 * The system is now under CVS at SourceForge (instead of the
3785 CVS repository on my home machine).
3786 * The new signal handling code has been tweaked to treat register
3787 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3788 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3789 downstream, e.g. in debug-int.lisp, has implicit dependencies
3790 on the unsignedness of integer representation of machine words,
3791 and that caused the system to bomb out with infinite regress
3792 when trying to recover from type errors involving signed values,
3793 e.g. (BUTLAST '(1 2 3) -1).)
3794 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3795 (This was one of the bugs Peter Van Eynde reported back in July.)
3796 * The system now uses code inspired by Colin Walters' O(N)
3797 implementation of MAP (from the cmucl-imp@cons.org mailing
3798 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3799 inline the MAP operation, and there is more than one
3800 sequence argument to the MAP call (so that it can't just
3801 do ETYPECASE once and for all based on the type of the
3802 single sequence argument). (The old non-inline implementation
3803 of the general M-argument sequence-of-length-N case required
3804 O(M*N*N) time when any of the sequence arguments were LISTs.)
3805 * The QUIT :UNIX-CODE keyword argument has been renamed to
3806 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3807 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3809 * Raymond Wiker's patches to port dynamic loading from Linux to
3810 FreeBSD have been added.
3811 * The BUGS file is now more nearly up to date, thanks in large part
3812 to Martin Atzmueller's review of it.
3813 * The debugger now flushes standard output streams before it begins
3814 its output ("debugger invoked" and so forth).
3815 * The core version number and fasl file version number have both
3816 been incremented, because of incompatible changes in the layout
3818 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3819 used to not be called for a saved Lisp image.)
3820 * Martin Atzmueller's version of a patch to fix a compiler crash,
3821 as posted on sbcl-devel 13 September 2000, has been installed.
3822 * Instead of installing Martin Atzmueller's patch for the
3823 compiler transform for SUBSEQ, I deleted the compiler transform,
3824 and transforms for some similar consing operations.
3825 * A bug in signal handling which kept TRACE from working on OpenBSD
3827 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3828 away by constant folding
3829 * The system now defines its address space constants in one place
3830 (in the Lisp sources), and propagates them automatically elsewhere
3831 (through GENESIS and the sbcl.h file). Therefore, patching the
3832 address map is less unnecessarily tedious and error-prone. The
3833 Lisp names of address space constants have also been systematized.
3834 * CVS tags like dollar-Header-dollar have been removed from
3835 the sources, because they have never saved me trouble and
3836 they've been source of trouble working with patches and other
3837 diff-related operations.
3838 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3839 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3841 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3843 * The system has been ported to OpenBSD.
3844 * The system now compiles with a simple "sh make.sh" on the systems
3845 that it's supported on. I.e., now you no longer need to tweak
3846 text in the target-features.lisp-expr and symlinks in src/runtime/
3847 by hand, the make.sh takes care of it for you.
3848 * The system is no longer so grossly inefficient when compiling code
3849 involving vectors implemented as general (not simple) vectors (VECTOR T),
3850 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3851 which dares to use the various sequence functions on non-simple
3852 vectors, takes less of a performance hit.
3853 * There is now a primitive type predicate VECTOR-T-P
3854 to test for the (VECTOR T) type, so that e.g.
3855 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3856 can now be compiled with some semblance of efficiency. (The old code
3857 turned the type declaration into a full call to %TYPEP at runtime!)
3858 * AREF on (VECTOR T) is still not fast, since it's still compiled
3859 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3860 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3861 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3862 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3863 without making full calls to SUBTYPEP at runtime.
3864 (Some analogous efficiency issues for non-simple vectors specialized to
3865 element types other than T, or for non-simple multidimensional arrays,
3866 have not been addressed. They could almost certainly be handled the
3867 same way if anyone is motivated to do so.)
3868 * The changes in array handling break binary compatibility, so
3869 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3870 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3871 instead of (VALUES T T).
3872 * By following the instructions that Dan Barlow posted to sbcl-devel
3873 on 2 July 2000, I was able to enable primitive dynamic object
3874 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3875 functionality is not implemented (since it calls ld to resolve
3876 library references automatically, requiring RUN-PROGRAM for its
3877 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3878 try to resolve library references) is now supported.
3879 * The system now flushes the standard output streams when it terminates,
3880 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3881 them at several other probably-convenient times, e.g. in each pass of
3882 the toplevel read-eval-print loop, and after evaluating a form given
3883 as an "--eval" command-line option. (These changes were motivated by a
3884 discussion of stream flushing issues on cmucl-imp in August 2000.)
3885 * The source transform for TYPEP of array types no longer assumes
3886 that an array whose element type is a not-yet-defined type
3887 is implemented as an array of T, but instead punts, so that the
3888 type will be interpreted at runtime.
3889 * There is now some support for cross-compiling in make.sh: each of
3890 the phases of make.sh has its own script. (This should be transparent
3891 to people doing ordinary, non-cross-compile builds.)
3892 * Since my laptop doesn't have hundreds of megabytes of memory like
3893 my desktop machine, I became more motivated to do some items on
3894 my to-do list in order to reduce the size of the system a little:
3895 ** Arrange for various needed-only-at-cold-init things to be
3896 uninterned after cold init. To support this, those things have
3897 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3898 symbols with such names are now uninterned after cold init).
3899 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3900 fasl files for cold load.
3901 ** Remove the old compiler structure pooling code (which used to
3902 be conditional on the target feature :SB-ALLOC) completely.
3903 ** Redo the representation of some data in cold init to be more compact.
3904 (I also looked into supporting byte compiled code at bootstrap time,
3905 which would probably reduce the size of the system a lot, but that
3906 looked too complicated, so I punted for now.)
3907 * The maximum signal nesting depth in the src/runtime/ support code has
3908 been reduced from 4096 to 256. (I don't know any reason for the very
3909 large old value. If the new smaller value turns out to break something,
3910 I'll probably just bump it back up.)
3911 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3913 * Many, many bugs reported by Peter Van Eynde have been added to
3914 the BUGS list; some have even been fixed.
3915 * While enabling dynamic object file loading, I tried to make the
3916 code easier to understand, renaming various functions and variables
3917 with less ambiguous names, and changing some function calling
3918 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3919 * While trying to figure out how to do the OpenBSD port, I tried to
3920 clean up some of the code in src/runtime/. In particular, I dropped
3921 support for non-POSIX signal handling, added various comments,
3922 tweaked the code to reduce the number of compilation warnings, and
3923 renamed some files to increase consistency.
3924 * To support the new automatic configuration functionality in make.sh,
3925 the source file target-features.lisp-expr has been replaced with the
3926 source file base-target-features.lisp-expr and the machine-generated
3927 file local-target-features.lisp-expr.
3928 * fixed a stupid quoting error in make.sh so that using CMU CL
3929 "lisp -batch" as cross-compilation host works again
3931 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3933 * DESCRIBE no longer tries to call itself recursively to describe
3934 bound/fbound values, so that it no longer fails on symbols which are
3935 bound to themselves (like keywords, T, and NIL).
3936 * DESCRIBE now works on generic functions.
3937 * The printer now prints less-screwed-up representations of closures
3938 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3939 * A private symbol is used instead of the :EMPTY keyword previously
3940 used to mark empty slots in hash tables. Thus
3941 (DEFVAR *HT* (MAKE-HASH-TABLE))
3942 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3943 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3944 now does what ANSI says that it should. (You can still get
3945 similar noncompliant behavior if bang on the hash table
3946 implementation with all the symbols you get back from
3947 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3948 This breaks binary compatibility, since tests for equality to
3949 :EMPTY are wired into things like the macroexpansion of
3950 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3952 * There's now a minimal placeholder implementation for CL:STEP,
3953 as required by ANSI.
3954 * An obscure bug in the interaction of the normal compiler, the byte
3955 compiler, inlining, and structure predicates has been patched
3956 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3957 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3959 * Missing ordinary arguments in a macro call are now detected even
3960 when the macro lambda list contains &KEY or &REST.
3961 * The debugger no longer complains about encountering the top of the
3962 stack when you type "FRAME 0" to explicitly instruct it to go to
3963 the top of the stack. And it now prints the frame you request even
3964 if it's the current frame (instead of saying "You are here.").
3965 * As specified by ANSI, the system now always prints keywords
3966 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3968 * The default initial SIZE of HASH-TABLEs is now smaller.
3969 * Type information from CLOS class dispatch is now propagated
3970 into DEFMETHOD bodies, so that e.g.
3971 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3973 is now basically equivalent to
3974 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3975 (DECLARE (TYPE SINGLE-FLOAT X))
3977 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3978 operation, without having to do run-time type dispatch.
3979 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3980 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3981 * Since I don't seem to be making big file reorganizations very often
3982 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3983 is overflowing my ability to conveniently back them up), I've finally
3984 checked the system into CVS. (The CVS repository is on my home system,
3985 not at SourceForge -- putting it on SourceForge might come later.)
3986 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3987 high-level GC-NOTIFY-FOO functions send their output. (There's
3988 still very little control of where low-level verbose GC functions
3989 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3990 less than it used to -- the GC-NOTIFY-FOO functions are now under
3991 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3992 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3993 in only one place in the source code, and propagates it automatically
3994 everywhere that it's needed. Thus e.g. when I bump the version from
3995 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3996 * The C source files now include boilerplate legalese and documentation
3997 at the head of each file (just as the Lisp source files already did).
3998 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3999 to his page will be replaced with a link to his new CLiki service.
4001 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4003 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4004 * The build process now looks for GNU make under the default name "gmake",
4005 instead of "make" as it used to. If GNU make is not available as "gmake"
4006 on your system, you can change this default behavior by setting the
4007 GNUMAKE environment variable.
4008 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4009 can build without error under CMU CL.
4011 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4013 * There is now a partial SBCL user manual (with some new text and some
4014 text cribbed from the CMU CL manual).
4015 * The beginnings of a profiler have been added (starting with the
4016 CMU CL profiler and simplifying and cleaning up). Eventually the
4017 main interface should be through the TRACE macro, but for now,
4018 it's still accessed through vaguely CMU-CL-style functions and macros
4019 exported from the package SB-PROFILE.
4020 * Some problems left over from porting CMU CL to the new
4021 cross-compilation bootstrap process have been cleaned up:
4022 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4023 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4024 part when getting rid of the compiler *BACKEND* stuff.)
4025 ** The value of *NULL-TYPE* was screwed up, because it was
4026 being initialized before the type system knew the final
4027 definition of the 'NULL type. This screwed up several key
4028 optimizations in the compiler, causing inefficiency in all sorts
4029 of places. (I found it because I wanted to understand why
4030 GET-INTERNAL-RUN-TIME was consing.)
4031 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4032 PROCLAIM FTYPE information. Unfortunately this broke binary
4033 compatibility again, since now the forms output by DEFGENERIC
4034 to refer to functions which didn't exist in 0.6.3.
4035 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4036 can use the new (as of 0.6.3) transform for SOME into MAP into
4038 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4039 values to (INTEGER 0 1000000), so that the time code will no longer
4040 occasionally get blown up by Linux returning 1000000 microseconds
4041 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4042 its output conform to the ANSI spec. (Alas, this makes its output
4043 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4044 * A full call to MAP NIL with a single sequence argument no longer conses.
4045 * fixes to problems pointed out by Martin Atzmueller:
4046 * The manual page no longer talks about multiprocessing as though
4047 it were currently supported.
4048 * The ILISP support patches have been removed from the distribution,
4049 because as of version 5.10.1, ILISP now supports SBCL without us
4050 having to maintain patches.
4051 * added a modified version of Raymond Toy's recent CMU CL patch for
4052 EQUALP comparison of HASH-TABLE
4054 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4056 * The system still can't cross-compile itself with
4057 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4058 entails), but at least it can compile more of itself that way
4059 than it used to be able to, and various buglets which were uncovered
4060 by trying to cross-compile itself that way have now been fixed.
4061 * This release breaks binary compatibility again. This time
4062 at least I've incremented the FASL file format version to 2, so that the
4063 problem can be detected reliably instead of just causing weird errors.
4064 * various new style warnings:
4065 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4066 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4067 ** using the lexical binding of a variable named in the *FOO* style
4068 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4069 as specified by ANSI.
4070 * *RANDOM-STATE* is no longer automatically initialized from
4071 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4072 default behavior of the system is to repeat its behavior every time
4073 it's run. If you'd like to change this behavior, you can always
4074 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4075 old convention there was no comparably easy way to get the system to
4076 repeat its behavior every time it was run.
4077 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4078 FTYPE declarations has been removed, in favor of their ANSI
4079 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4080 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4081 the types of their sequence arguments can be determined at compile time.
4082 This is done through a new open code expansion for MAP which eliminates
4083 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4084 argument types can be determined at compile time.
4085 * The optimizer now transforms COERCE into an identity operation when it
4086 can prove that the coerced object is already of the correct type. (This
4087 can be a win for machine generated code, including the output of other
4088 optimization transforms, such as the MAP transform above.)
4089 * Credit information has been moved from source file headers into CREDITS.
4090 * Source file headers have been made more standard.
4091 * The CASE macro now compiles without complaining even when it has
4094 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4096 * (Note that the way that the PCL macroexpansions were rewritten
4097 to accommodate the change in DEFGENERIC below breaks binary
4098 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4099 not run under sbcl-0.6.2. Once we get out of alpha releases,
4100 i.e. hit release 1.0.0, we'll probably try to maintain binary
4101 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4102 and sbcl-1.4.4. Until then, however, it might be fairly common
4103 for maintenance releases to break binary compatibility.)
4104 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4106 * The reporting of unhandled signals has been changed to print some
4107 explanatory text as well as the report form. (Previously only
4108 the report form was printed.)
4109 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4110 it defines, so that the compiler no longer issues undefined function
4111 warnings for compiled-but-not-yet-loaded generic functions.
4112 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4113 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4114 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4115 section of the ANSI spec are supported.
4116 * The "" nickname for the "KEYWORD" package has been removed.
4117 The reader still handles symbol tokens which begin with a package marker
4118 as keywords, but it doesn't expose its mechanism for doing so in the
4119 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4120 * The system now issues STYLE-WARNINGs for contradictory TYPE
4121 proclamations. (Warnings for contradictory FTYPE proclamations would
4122 be nice too, but those can't be done usefully unless the type system
4123 is made smarter about FUNCTION types.)
4124 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4125 systematized, so that "*target-*.lisp is supposed to exist only on the
4126 target and imply that there's a related file which exists on the
4127 host, and *host-*.lisp is supposed to exist only on the host and imply
4128 that there's a related file which exists on the target. This involves a
4129 lot of renaming. Hopefully the acute confusion caused by the renaming
4130 will be justified by the reduction in chronic confusion..
4131 ** runtime-type.lisp -> early-target-type.lisp
4132 ** target-type.lisp -> late-target-type.lisp
4133 ** early-host-format.lisp -> early-format.lisp
4134 ** late-host-format.lisp -> late-format.lisp
4135 ** host-error.lisp -> misc-error.lisp
4136 ** early-error.lisp -> early-target-error.lisp
4137 ** late-error.lisp -> late-target-error.lisp
4138 ** host-defboot.lisp -> early-defboot.lisp
4139 ** code/misc.lisp -> code/target-misc.lisp
4140 ** code/host-misc.lisp -> code/misc.lisp
4141 ** code/numbers.lisp -> code/target-numbers.lisp
4142 ** code/early-numbers.lisp -> numbers.lisp
4143 ** early-host-type.lisp -> early-type.lisp
4144 ** late-host-type.lisp -> late-type.lisp
4145 ** host-typep.lisp -> typep.lisp
4146 ** load.lisp -> target-load.lisp
4147 ** host-load.lisp -> load.lisp
4148 ** host-disassem.lisp -> disassem.lisp
4149 ** host-insts.lisp -> insts.lisp
4150 ** byte-comp.lisp -> target-byte-comp.lisp
4151 ** host-byte-comp.lisp -> byte-comp.lisp
4152 ** host-signal.lisp -> signal.lisp
4153 ** host-defstruct.lisp -> defstruct.lisp
4154 ** late-target-type.lisp -> deftypes-for-target.lisp
4155 Furthermore, several other previously target-only files foo.lisp (e.g.
4156 hash-table.lisp and random.lisp) have been split into a target-and-host
4157 foo.lisp file and a target-only target-foo.lisp file, with their key type
4158 definitions in the target-and-host part, so that the cross-compiler will
4159 know more about target types.
4160 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4161 gone away. In their place are various *BACKEND-FOO* variables
4162 corresponding to the slots of the old structure.
4163 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4164 into the target SBCL has been fixed.
4165 * The chill.lisp system for loading cold code into a running SBCL
4167 * Support for the CMU CL "scavenger hook" extension has been removed.
4168 (It was undocumented and unused in the CMU CL sources that SBCL was
4169 derived from, and stale in sbcl-0.6.1.)
4170 * Various errors in the cross-compiler type system were detected
4171 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4172 (enabling various consistency checks). Many of them were fixed,
4173 but some hard problems remain, so the compiler is back to
4174 running without *TYPE-SYSTEM-INITIALIZED* for now.
4175 * As part of the cross-compiler type system cleanup, I implemented
4176 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4177 * I have started adding UNCROSS calls throughout the type system
4178 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4179 on cross-compiler input files will be able to go away, and various
4181 * CONSTANTP now returns true for quoted forms (as explicitly required
4184 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4186 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4187 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4188 (MAKE-LIST -1). In the longer term, it ought to become true
4189 that declarations are assertions even at SAFETY 1. For now, it's not
4190 quite true even at SAFETY 3, but it's at least more nearly true..
4191 (Note that this change seems to increases the size of the system by
4192 O(5%) and to decrease the speed of the compiler by 20% or more.)
4193 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4194 for the problem of printing dozens of lines of distracting information
4195 about low-level system machinery as part of the top stack frame
4196 on entry to the debugger when an undefined function was called.
4197 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4198 is not reset to COMMON-LISP-USER.
4199 * Compilation of stuff related to dyncount.lisp has been made conditional
4200 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4201 smaller. The various dyncount-related symbols have been moved into
4202 a new "SB-DYNCOUNT" package.
4203 * tty-inspect.lisp has been renamed to inspect.lisp.
4204 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4205 feature has gone away. (When we eventually port to other flavors of
4206 libc and/or Unix, we'll try to make the differences between flavors
4207 invisible at the user level.)
4208 * Various other *FEATURES* tags, and/or their associated conditionals,
4209 have been removed if obsolescent, or given better documentation, or
4210 sometimes given more-mnemonic names.
4212 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4214 * tidied up "make.sh" script
4215 * tidied up system directory structure
4216 * better "clean.sh" behavior
4217 * added doc/FOR-CMUCL-DEVELOPERS
4218 * many many small tweaks to output format, e.g. removing possibly-confusing
4219 trailing #\. character in DESCRIBE-INSTANCE
4220 * (EQUALP #\A 'A) no longer signals an error.
4221 * new hashing code, including EQUALP hashing
4222 * tidied up Lisp initialization and toplevel
4223 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4224 * command line argument processing
4225 * added POSIX-GETENV function to deal with Unix-ish environment variables
4226 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4227 terminating SBCL on EOF
4228 * non-verbose GC by default
4229 * There is no more "sbcl" shell script; the sbcl file is now the C
4230 runtime executable (just like CMU CL).
4231 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4233 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4234 symbol and package information as Lisp native symbol and package objects
4235 instead of strings naming symbols and strings naming packages. This way,
4236 whenever packages are renamed (as in warm init), debug information is
4237 transformed along with everything else.
4238 * tweaked the optimization policy declarations which control the building
4239 of SBCL itself. Now, among other things, the system no longer saves
4240 source location debugging information. (This helps two problems at once
4241 by reducing SBCL size and by keeping SBCL from trying to look for its
4242 sources -- which may not exist -- when reporting errors.)
4243 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4244 debugging and testing purposes
4245 * cleaned up printing, making the printer call PRINT-OBJECT for
4246 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4247 methods, giving nearly-ANSI behavior
4248 * converted almost all special variables to use *FOO* naming convention
4249 * deleted PARSE-TIME functionality, since it can be done portably
4250 * moved some files out of cold init into warm init
4251 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4253 * regularized formatting of source files
4254 * added an install.sh script
4255 * fixed ridiculous memory usage of cross-compiler by making
4256 compiler/alloc.lisp not try to do pooling unless it can hook
4257 itself into the GC of the cross-compilation host. Now the system
4258 builds nicely on my old laptop.
4259 * added :SB-ALLOC in target-features.lisp-expr
4260 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4261 was not implemented)
4262 * re-did condition handling and note reporting in the compiler. Notes
4263 are no longer handled by signalling conditions. Style warnings
4264 and warnings are handled more correctly and reported in such a way
4265 that it's easy to find one or the other in your output (so that you
4266 can e.g. figure out which of many problems caused COMPILE-FILE to
4268 * changed the severity of several compiler warnings from full WARNING
4269 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4270 compiler note reporting so that it doesn't use the condition system
4271 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4272 COMPILE-FILE command)
4273 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4274 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4275 no longer supported, since their implementation was deeply intertwingled
4276 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4277 * removed bogus "support" for compiler macros named (SETF FOO), and
4278 removed the compiler macro for SETF INFO (but only after making a fool
4279 of myself on the cmucl-imp mailing list by posting a bogus patch for
4280 DEFINE-COMPILER-MACRO..)
4281 * Compiled files containing forms which have side effects on the Lisp
4282 reader (such as DEFPACKAGE forms) are now handled more correctly.
4283 (Compiler queuing of top level lambdas has been suppressed by setting
4284 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4285 * deleted various currently-unused source files, e.g. gengc.lisp. They
4286 may be added back at some point e.g. when porting to other architectures,
4287 but until they are it's distracting to distribute them and to try to
4289 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4290 there were so many of them they're just distractions, and UNCROSS is
4291 known to be able to handle the current sources
4292 * moved PROFILE functionality into TRACE, so that it will be clear
4293 how the wrapping and unwrapping of functions when you profile them
4294 interacts with the wrapping and unwrapping of functions when you
4295 trace them. (Actually, the functionality isn't there yet, but at least
4296 the interface specification is there. Hopefully, the functionality will
4297 arrive with some future maintenance release.)
4298 * removed host-oops.lisp
4299 * changed signature of QUIT function to allow UNIX-CODE argument
4300 * fixed READ-SEQUENCE bug
4301 * tweaked verbose GC output so that it looks more like the progress
4302 output that ANSI specifies for functions like LOAD
4303 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4304 * added <http://sbcl.sourceforge.com> to the banner information printed by