1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
3 * MOP improvement: support for user-defined subclasses of
4 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
5 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
6 the expansion of DEFMETHOD, is responsible for generating a form
7 which creates a list of specializers when evaluated. Additional
8 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
9 debugging and introspective support.
10 * minor incompatible change: the (unsupported) spinlock interface
11 has changed: free spinlock now has the value NIL, and a held spinlock
12 has the owning thread as its value.
13 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
14 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
16 * enhancement: name of a socket-stream is now "a socket" instead of
18 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
19 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
20 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
21 for generic functions. (Thanks to Tobias C. Rittweiler)
22 * optimization: bignum printing speed has been improved by 20-40%
23 (depending on the bignum size.)
24 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
26 * bug fix: the cache used by the CLOS to store precomputed effective
27 methods, slot offsets, and constant return values is now thread and
29 * bug fix: generic function dispatch function updating is now thread
30 and interrupt safe (in the sense that the known issues have been
32 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
33 * bug fix: interning EQL-specializers is now thread and interrupt safe.
34 * bug fix: asdf systems with dependencies to the SB-POSIX or
35 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
37 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
38 * new contrib: sb-cover, an experimental code coverage tool, is included
40 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
42 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
43 produce more efficient code on x86-64 if the compiler can determine
44 that (+ INDEX <constant>) does not require a bounds check and FOO
45 has an element type at least 8 bits wide.
46 * enhancement: a new, experimental synchronous timeout facility is
47 provided. Refer to SB-SYS:WITH-DEADLINE for details.
48 * enhancement: when a symbol name conflict error arises, the
49 conflicting symbols are always printed with a package prefix.
50 (thanks to Kevin Reid)
51 * enhancement: stepping is now once again supported on the SPARC and
52 MIPS platforms. (It is also now more likely to work on CheneyGC
54 * enhancement: sb-sprof can now also track and report accurate call
56 * bug fixes: the treatment of non-standard subclasses of
57 SB-MOP:SPECIALIZER is more correct.
58 * incompatible change: PURIFY no longer copies the data from the
59 dynamic space into the static and read-only spaces on platforms
60 that use the generational garbage collector
61 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
63 * bug fix: GC race condition occasionally resulting in crashes with
64 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
66 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
68 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
69 a STYLE-WARNING to be signalled (regression from 1.0.4.)
70 * bug fix: an asynchronous interrupt could previously leave the
71 system running with GC inhibited.
72 * bug fix: a DECLARE form evaluated at top-level now causes an error
73 rather than silently (or verbosely) returning NIL.
74 * bug fix: trying to dissassemble functions compiled at high DEBUG could
75 result in a error being signalled due to source form lookup errors.
76 (reported by Peter Graves)
78 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
79 * incompatible change: removed writer methods for host-ent-name,
80 host-ent-addresses -- changing the values did not update the DNS
82 * minor incompatible change: changed experimental JOIN-THREAD interface
83 * documentation: the manual now lists reader and writer methods
84 in class slot documentation sections. (thanks to Richard M Kreuter)
85 * documentation: unwinding from asyncronous events has been
87 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
88 in multithreaded application code.
89 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
90 platforms. (thanks to James Anderson for the optimization hint)
91 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
93 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
94 variants no longer cons.
95 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
96 their NOT- variants no longer cons.
97 * optimization: EQUAL hash tables no longer use SXHASH for objects
98 of all data types, but instead use an EQL hash for types for which
99 EQUAL is the same as EQL.
100 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
101 are significantly faster.
102 * optimization: new STRING-OUTPUT-STREAM implementation is much
103 faster for large outputs and conses 30% less on average.
104 * enhancement: XREF information is now collected for references made
105 to global variables using SYMBOL-VALUE and a constant argument.
106 * enhancement: SIGINT now causes a specific condition
107 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
108 * bug fix: FILE-POSITION used to signal an error in some cases where
109 ANSI requires it to return NIL.
110 * bug fix: ADJUST-ARRAY is now interrupt-safe.
111 * bug fix: adding and removing fd-handlers is now interrupt-safe.
112 * bug fix: inlined calls to C now ensure 16byte stack alignment on
114 * bug fix: bad type declaration in the CLOS implementation has
115 been fixed. (reported by James Anderson)
116 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
117 (thanks to Lutz Euler)
118 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
119 required by the CLHS. (reported by Samium Gromoff)
120 * bug fix: dead unbound variable references now signal an error.
121 (reported by Marco Monteiro)
122 * bug fix: / with an unused value was being deleted in safe code.
123 (thanks to Marco Monteiro and Kevin Reid)
124 * bug fix: number of characters that can be written onto a single
125 line in a file is unlimited.
126 * bug fix: some GC deadlocks caused by asynchronous interrupts have
127 been fixed by inhibiting interrupts for when GC is disbled.
128 * bug fix: some interrupt safety issues with GETHASH, (SETF
129 GETHASH), CLRHASH and REMHASH have been fixed.
130 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
131 specifier no longer causes infinite recursion.
132 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
133 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
134 * bug fix: an &environment argument with the correct variable information
135 is passed to macros that are expanded during byte compilation.
136 (reported by Samium Gromoff)
137 * bug fix: a logic error was causing the hash function for bignums to
138 have excessive amounts of collisions. (reported by Faré Rideau)
139 * bug fix: modifying the contents of an array could change the return
140 value of SXHASH on that array, which is only allowed for strings
141 and bit vectors. (bug introduced in 0.9.16)
142 * improvement: the x86-64/darwin port now passes all tests (except
143 for the debugger tests) but should still be considered
144 experimental until this is fixed.
145 * improvement: a style-warning is signaled for CASE (etc) clauses with
146 duplicate keys. (thanks to Kevin Reid)
147 * improvement: macos/x86 and macos/x86-64 now use mach exception
148 handlers for dealing with illegal instructions (for trapping and
149 error handling) and memory protection violations (for GC).
151 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
152 * new platform: experimental support for x86-64/darwin (MacOS).
153 * incompatible change: the thread-safe (on most platforms) getaddrinfo
154 and getnameinfo sockets functions are used instead of gethostbyaddr
155 and gethostbyname, on platforms where the newer functions are available.
156 As a result, the ALIASES field of HOST-ENT will always be NIL on these
158 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
159 don't print the startup banner, but behave as if --noinform was passed
160 as a command line argument. (thanks to Kevin Reid)
161 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
162 * optimization: code using alien values with undeclared types is much faster.
163 * optimization: the compiler is now able to open code SEARCH in more cases.
164 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
165 * bug fix: using standardized COMMON-LISP special variables as loop
166 variables no longer signals bogus package lock violations. (reported
168 * bug fix: declaring local loop variables to be of a range-limited type
169 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
170 (reported by Andras Simon)
171 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
172 bugs remain on x86-64.)
173 * bug fix: the #= and ## reader macros now interact reasonably with
174 funcallable instances.
175 * bug fix: type-checks for function arguments were compiled using the
176 compiler policy settings of the wrong lexical environment (in 1.0.2
178 * bug fix: SHADOW accepts characters as string designators, as required
179 by the spec (thanks to Eric Marsden)
180 * bug fix: fixed GC safety issues when foreign functions are called with
181 non-base strings as arguments
182 * bug fix: more consistent error output for fatal-compiler-warnings, like
184 * bug fix: functions with proper names were showing up as NIL in
185 backtraces (reported by Edi Weitz, regression in 1.0.2)
187 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
188 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
189 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
190 produce more efficient code on the x86 if the compiler can determine
191 that (+ INDEX <constant>) does not require a bounds check and FOO
192 has an element type at least 8 bits wide.
193 * bug fix: references to undefined variables are handled the same way
194 in toplevel forms as in the normal compiler
195 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
197 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
198 (thanks to Jon Buller)
199 * improvement: faster compilation times for complex functions
200 * improvement: added readlink support to SB-POSIX (thanks to Richard
203 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
204 * improvement: experimental support for mach exception handling on
205 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
207 * improvement: support for GBK external format.
208 (thanks to Chun Tian (binghe))
209 * improvement: the debugger now displays variables that have been closed
210 over, in code compiled with (DEBUG 2) or higher
211 * improvement: support for executable cores on NetBSD (thanks to
213 * new feature: added a RESTART-FRAME debugger command
214 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
215 be used to provide an implementation for FILE-POSITION on Gray streams
216 (thanks to Eric Marsden)
217 * new feature: add syslog, openlog and closelog support to SB-POSIX
218 (thanks to Richard Kreuter)
219 * optimization: the function call overhead in code compiled with
220 a high DEBUG optimization setting is significantly
221 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
222 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
223 (thanks to Tony Martinez)
224 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
225 evaluated once (reported by Kevin Reid)
226 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
227 works again (thanks to Kilian Sprotte)
228 * bug fix: an error is signaled for tagbodies with duplicate tags
229 (thanks to Stephen Wilson)
230 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
231 * bug fix: Win32 port can now handle foreign code unwinding Lisp
232 stack frames from alien callbacks.
233 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
234 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
235 65k pages (thanks to David Woodhouse)
236 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
238 changes in sbcl-1.0.1 relative to sbcl-1.0:
239 * new platform: FreeBSD/x86-64, including support for threading.
240 * new feature: the compiler stores cross-referencing information
241 abount function calls (who-calls), macroexpansion (who-macroexpands)
242 and special variables (who-binds, who-sets, who-references) for code
243 compiled with (< SPACE 3). This information is available through the
244 sb-introspect contrib.
245 * new feature: users may subclass SEQUENCE, and have instances of
246 these classes interoperate with standard Common Lisp functions if
247 a number of methods are defined. (This feature is experimental
248 and the interface subject to change based on feedback from SBCL
249 users and the general community)
250 * improvement: sb-sprof traces call stacks to an arbitrary depth on
251 x86/x86-64, rather than the previous fixed depth of 8
252 * improvement: another pthread back-end of mutex "pthread-futex".
253 * bug fix: non-ascii command-line arguments are processed correctly
254 (thanks to Yaroslav Kavenchuk)
255 * bug fix: non-required arguments were not passed correctly when a method
256 defined using DEFMETHOD was called from a mop-generated method using
257 CALL-NEXT-METHOD (reported by Pascal Costanza)
258 * bug fix: recursion is now permitted in accessors through
259 SLOT-UNBOUND. (reported by Pascal Costanza)
260 * bug fix: an error was signaled at startup if the HOME environment
261 variable was defined, but had an empty value (reported by Peter Van Eynde)
262 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
263 signal a PROGRAM-ERROR, not a BUG.
264 * bug fix: SB-POSIX:READDIR works when built with large file support.
265 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
267 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
268 for better automatic indentation support. (thanks to Matt Pillsbury)
269 * optimization: loading generic functions no longer takes O(n^2) time,
270 proportional to the amount of methods in the generic function
271 (reported by Todd Sabin and Jeremy Brown)
272 * optimization: the FIND and POSITION family of sequence functions
273 are significantly faster on arrays whose element types have been
275 * improvements to the Windows port:
276 ** Intermittent heap corruption problems have been fixed. (thanks
277 to Alastair Bridgewater)
278 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
280 ** Lisp is able to unwind foreign exception frames from alien
281 callbacks. (thanks to Alastair Bridgewater)
283 changes in sbcl-1.0 relative to sbcl-0.9.18:
284 * improvement: experimental support for threading on FreeBSD/x86.
285 (thanks to NIIMI Satoshi)
286 * improvement: runtime option --dynamic-space-size can be used
287 to set the size of the dynamic space reserved on startup.
288 * improvement: floating point modes in effect are now saved in
289 core, and restored on startup.
290 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
291 startup, not time since first call to GET-INTERNAL-REAL-TIME.
292 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
293 threads are not running after *SAVE-HOOKS* have run.
294 * improvement: writes to CLOS instance slots are type-checked in code
295 compiled with (SAFETY 3)
296 * improvement: floating-point exception handling on FreeBSD (thanks to
298 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
299 (thanks to Zach Beane)
300 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
302 * improvement: added support for the Shift-JIS external format.
303 (contributed by NIIMI Satoshi)
304 * improvement: callbacks are supported on Linux/PPC. (thanks to
306 * bug fix: compiler bug triggered by a (non-standard) VALUES
307 declaration in a LET* was fixed. (reported by Kaersten Poeck)
308 * bug fix: file compiler no longer confuses validated and already
309 dumped structurres. (reported by Kaersten Poeck)
310 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
311 fill-pointer signals a type-error as required. (thanks to
313 * bug fix: disassemly of funcallable instances works.
314 * bug fix: single stepping on PPC.
315 * bug fix: fix thread-safety problems in the type system (generally
316 manifesting as nonsensical errors like "STRING is a bad type specifier
317 for sequences" or "The value 1 is not of type FIXNUM")
318 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
319 (reported by Josip Gracin)
320 * bug fix: an error is signaled for attempts to displace arrays with
321 incompatible element types (thanks to Mario Mommer)
322 * bug fix: more correct handling of wide characters in debug info
323 (bug reported by Attila Lendvai and fixed by Juho Snellman)
324 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
325 and don't cause extra consing
326 * optimization: MAP and MAP-INTO are significantly faster on vectors
327 whose elements types have been declared.
328 * Improvements to SB-SPROF:
329 ** Support for allocation profiling
330 ** Reduced profiling overhead, especially for long profiling runs
331 * Improvements to the Windows port:
332 ** floating point exceptions are now reported correctly.
333 ** stack exhaustion detection works partially.
334 ** more accurate GET-INTERNAL-REAL-TIME.
335 ** better RUN-PROGRAM behaviour in case of errors in the child
337 ** PROBE-FILE now simplifies pathnames correctly.
338 ** DIRECTORY now works correctly with :WILD-INFERIORS.
339 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
340 1st 1970 (although time timezone data will be unreliable).
341 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
343 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
344 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
345 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
347 * bug fix: two potential GC deadlocks affecting threaded builds.
348 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
349 returns T (reported by Anton Kazennikov)
350 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
351 with non-variable places
352 * bug fix: remove a race condition in the setting of
353 funcallable-instance functions, this should make threaded CLOS
354 code more stable against memory faults.
355 * bug fix: corruption of specials when unbinding is interrupted by an
356 asynchronous unwind (reported by Hannu Koivisto)
357 * improvement: the debugger will now also display local variables that
358 are only used once, for code compiled with a DEBUG optimization quality
361 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
362 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
363 * incompatible change: External-format support for FFI calls. The
364 SB-ALIEN:C-STRING no longer implies an ASCII
365 external-format. Instead the string is subject to external-format
366 conversion. Additionally return values of type SB-ALIEN:C-STRING
367 are by default of type (SIMPLE-ARRAY CHARACTER), not
368 SIMPLE-BASE-STRING. For an alien type that behaves like the old
369 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
370 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
371 * incompatible change: SB-EXT package no longer contains the
372 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
373 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
375 * incompatible change: the single-stepper is no longer available
376 on Alpha, Mips and Sparc platforms.
377 * minor incompatible change: the direct superclasses of
378 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
379 not (STANDARD-OBJECT FUNCTION). This makes the
380 class-precedence-lists of GENERIC-FUNCTION and
381 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
383 * minor incompatible change: the default stream external format on
384 non-unicode SBCL is no longer detected from the locale, but is always
386 * new feature: Add a version of evaluator that uses an interpreter instead
387 of the compiler. EVAL still uses the compiler by default, to switch it
388 to use the interpreter, set the value of the variable
389 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
390 * minor incompatible change: the single-stepper REPL has been merged
391 with the normal debugger (see the "Stepping" heading of the debugger help
392 for more details). The debugger command STEP will no longer switch
393 to the single-stepper REPL.
394 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
395 for a type now works.
396 * bug fix: loading of default sysinit file works. (thanks to Leonid
398 * bug fix: better detection of circularities in the file-compiler.
399 (reported by Marco Monteiro)
400 * bug fix: the CL pathname functions now work with files that have
401 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
402 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
403 systems with Unix98 pty semantics.
404 * bug fix: ASDF-INSTALL will now work with bsd tar.
405 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
407 * bug fix: timers expiring in dead threads no longer cause a
408 type-error (reported by Paul "Nonny Mouse").
409 * bug fix: thanks to more lightweight single-stepper instrumentation,
410 code compiled with (DEBUG 3) will compile and execute significantly faster,
411 and will have more accurate type-inferencing than before
412 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
413 whose bindings are modified
414 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
415 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
416 ** replace dummy implementations of CL:MACHINE-INSTANCE and
417 CL:SOFTWARE-VERSION with proper version
419 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
420 * feature: implemented the READER-METHOD-CLASS and
421 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
422 as specified by AMOP.
423 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
425 * optimization: faster LOGCOUNT implementation on x86 and x86-64
426 (thanks to Lutz Euler)
427 * optimization: hashing of general arrays and vectors has been
428 improved. (reported by Any Fingerhut)
429 * enhancement: SB-INTROSPECT is now able to find definitions of
430 profiled functions. (thanks to Troels Henriksen)
431 * enhancement: compiler-macro expansion applies now to FUNCALL forms
432 as well, allowing compiler-macros for SETF-functions to expand.
433 * enhancement: step-instrumentation no longer wraps known
434 single-value functions with multiple-value context, allowing
435 better type inference.
436 * fixed bug #337: use of MAKE-METHOD in method combination now works
437 even in the presence of user-defined method classes. (reported by
438 Bruno Haible and Pascal Costanza)
439 * fixed bug #339(c): if there are applicable methods not part of any
440 long-form method-combination group, call INVALID-METHOD-ERROR.
441 (reported by Bruno Haible)
442 * fixed bug #361: the :FUNCTION initarg in the protocol for
443 initialization of methods can now be used to override
444 internally-produced optimized functions. (reported by Bruno
446 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
447 system-provided lambda expression no longer cause warnings about
448 unbound #:|pv-table| symbols.
449 * bug fix: improved the handling of type declarations and the
450 detection of violations for keyword arguments with non-constant
452 * bug fix: potentially erronous calls to PATHNAME and
453 MERGE-PATHNAMES were being flushed in some cases.
454 (reported by Richard Kreuter)
455 * bug fix: compiled calls to TYPEP were mishandling obsolete
456 instances. (reported by James Bielman and Attila Lendvai)
457 * bug fix: format strings with ~> without matching ~< no longer
458 trigger an AVER, but signal an understandable error instead.
459 (reported by Antonio Martinez)
460 * bug fix: specifying an output-file without a pathname-type for
461 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
462 of input-file instead of "fasl". (reported by Robert Dodier)
463 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
464 (reported by James Y Knight).
465 * bug fix: compiler-macros-function did not consider the environment
466 argument for shadowing by local functions.
467 * bug fix: compiler-macros expansion was inhibited by local INLINE
469 * bug fix: inline expansions of known functions were subject to
470 step-instrumentation in high debug policies, leading to problems
472 * bug fix: compiler failed to differentiate between different CONS
474 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
475 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
476 * bug fix: #368: incorrect use of expressed vs. upgraded array
478 * bug fix: #306a: more precise unions of array types.
479 * thread-safety improvements:
480 ** CONDITION-WAIT could return early on Linux, if the thread was
481 interrupted and subsequently continued with SIGCONT.
482 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
484 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
485 * added support for the ucs-2 external format. (contributed by Ivan
487 * minor incompatible change: pretty printing of objects of type
488 (cons symbol) is, in the default pprint-dispatch-table, now
489 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
491 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
492 called later in a class's lifetime, possibly as late as when the
493 first instance of the class is created. Previously,
494 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
495 class became finalizeable.
496 * fixed bug: FILE-POSITION sometimes returned inconsistent results
497 for multibyte external-format streams. (thanks to "vbzoli")
498 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
499 with :ALLOCATION :CLASS inherited from superclasses of the
501 * fixed bug: anonymous classes can now be created using the :NAME
502 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
503 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
504 * fixed bug: core-files saved with :EXECUTABLE T can again be
505 executed when SBCL_HOME isn't set. (reported by James Knight)
506 * fixed bug: toplevel LOCALLY forms with declarations could
507 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
508 * fixed bug: printing from several different threads using different
509 values of *print-case* could cause invalid output, due to
510 some internal special variables of the printer not being bound
511 thread-locally (reported by Max Mikhanosha)
512 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
513 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
514 (reported by Pascal Costanza)
515 * fixed bug: FINALIZE-INHERITANCE is called from
516 REINITIALIZE-INSTANCE on classes when the class has previously
517 been finalized, as required by AMOP.
518 * minor code generation optimizations:
519 ** better register allocation in CLOS dispatching functions
520 ** overflow detection when coercing signed bytes to fixnums on x86-64
521 ** is now implemented with one IMUL instruction instead of three shifts
522 ** more efficient bit-vector access on x86 and x86-64
523 ** more efficient access to raw structure slots on x86 and x86-64
524 * fixed some bugs revealed by Paul Dietz' test suite:
525 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
528 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
529 * feature: thread support on Solaris/x86, and experimental thread support
531 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
533 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
534 default method for SLOT-UNBOUND.
535 * minor incompatible change: prevent the user from specializing the
536 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
537 somewhat counter to the intent of the protocol, I (CSR) think, and
538 additionally it just doesn't work in SBCL as currently
539 implemented, thanks to optimizations (that are always valid for
540 the other three SLOT-VALUEish functions, but not for the setter).
541 * minor incompatible change: the :SB-LDB feature is now enabled by
542 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
543 the low-level debugger.
544 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
545 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
546 down to an SB-SYS:FD-STREAM, allowing direct communication in
547 more cases. Temporary files and pipes are still used as a fallback
549 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
550 on the non-reentrant C functions localtime(3) and gmtime(3).
551 * bug fix: global optimization policy was not visible in LOCALLY and
553 * bug fix: class objects can be used as specializers in methods.
554 (reported by Pascal Costanza)
555 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
556 with a type error. (reported by blitz_ on #lisp)
557 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
558 failed with a type error. (reported by Pascal Bourguignon)
559 * bug fix: merging pathnames against defaults with :DIRECTORY
560 starting with '(:RELATIVE :BACK) should preserve the :BACK.
561 (reported by James Y Knight)
562 * bug fix: saving large (>2GB) cores on x86-64 now works
563 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
564 constant index and a (simple-array (signed-byte 32)) array
565 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
566 error (patch by Robert J. Macomber)
567 * feature: TIME now displays the amount of run-time spent in GC
568 * bug fix: The debugger now does a better job of respecting
569 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
570 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
571 workaround for bug 403.)
572 * bug fix: floating point exception handling now works on FreeBSD
573 (thanks to NIIMI Satoshi)
574 * fixed some bugs revealed by Paul Dietz' test suite:
575 ** MISC.641: LET-conversion were not supposed to work in late
577 * improvements to the Win32/x86 port:
578 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
579 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
581 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
582 * new feature: source path information is generated for macro-expansion
583 errors for use in IDE's like Slime (thanks to Helmut Eller)
584 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
585 * bug fix: compiler does not loop forever on an invalid type in
587 * improvement: compilation of most CLOS applications is significantly
589 * optimization: added a limited bytecode compiler for simple toplevel
590 forms, speeding up compilation and FASL loading
591 * bug fix: the statistical profiler now properly distinguishes anonymous
594 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
595 * minor incompatible change: in sbcl-0.9.11 (but not earlier
596 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
597 into subclasses of the finalized class. Now user calls to
598 FINALIZE-INHERITANCE finalize just the one class, and calls by the
599 system return before any subclasses are finalized.
600 * minor incompatible change: The reader no longer ignores errors
601 regarding non-existent packages in #+ and #- feature tests.
602 * new feature: command line options --no-sysinit, --no-userinit to
603 inhibit loading the corresponding init files
604 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
605 for sending data through UDP sockets (thanks to François-René Rideau)
606 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
607 error is signalled instead (thanks to François-René Rideau)
608 * minor incompatible change: Error signalling behaviour of lexical
609 operations violating package locks has changed slightly. Refer to
610 documentation on package locks for details.
611 * enhancement: EVAL can process IF-expressions without resorting to the
613 * bug fix: Some operations involving SIN, COS, or TAN weren't being
614 constant-folded properly due to missing out-of-line functions.
615 (reported by Mika Pihlajamäki)
616 * bug fix: LISTEN sometimes returned T even in cases where no data was
617 immediately available from the stream
618 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
619 were reversed. (reported by Levente Mészáros)
620 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
621 when given a vector displaced to an adjustable array. (reported
623 * bug fix: derivation of float boundaries from numbers outside the
624 appropriate float range (reported by John Wiseman)
625 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
627 * bug fix: precomputing cache entries for generic functions with
628 some subclasses of specializers as yet invalid does not attempt to
629 fill a cache line with a negative offset. (reported by Levente
631 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
632 allow condition class objects as arguments to DOCUMENTATION and
633 (SETF DOCUMENTATION); only find and set documentation for
634 structure names for the STRUCTURE doc-type. (suggested by Gary
636 * improvements to the Win32/x86 port:
637 ** (user-homedir-pathname) and default initialization file
638 locations now know about the user's "Documents and Settings"
639 directory (thanks to Yaroslav Kavenchuk)
640 ** run-program is implemented (thanks to Mike Thomas)
641 ** sockets support (thanks to Timothy Ritchey)
642 ** better backtrace support (thanks to Alastair Bridgewater)
643 ** sb-grovel supported
644 ** asdf-install and sb-posix work somewhat
645 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
646 * improvements to the Solaris/x86 port:
647 ** works on Solaris 11/Solaris Express
648 ** floating-point exception handling support
649 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
650 * fixed some bugs revealed by Paul Dietz' test suite:
651 ** REMOVE-METHOD returns its generic function argument even when
652 the method is not one of the generic functions' methods.
653 ** objects declared as MEMBER types can be admissible arguments to
655 ** printing characters should simply be printed by the FORMAT ~:C
657 ** compiler failure when compiling functions with hairy constant
658 defaults for optional parameters.
659 ** compiler produces wrong code when MAYBE-INLINE-expanding a
660 function, which is already optimized.
662 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
663 * new platform: experimental support for SBCL x86/Darwin, including
664 MacOS X 10.4.5 on Intel.
665 * new feature: Unicode character names are now known to the system
666 (through CHAR-NAME and NAME-CHAR).
667 * new feature: the filesystem location of SBCL's core file is
668 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
669 * minor incompatible change: the contrib modules SB-POSIX and
670 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
671 this change is to make it easier to distribute
672 location-independent binaries.
673 * bug fix: as implied by AMOP, standardized classes no longer have
674 slots named by external symbols of public packages. (reported by
676 * optimization: calling functions via constant symbols -- as in
677 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
678 via the function object as in (FUNCALL #'FOO).
679 * optimization: on x86 and x86-64, the calling convention for the
680 case of dealing with an unknown number of values has been altered
681 to be friendlier to the prediction heuristics implemented,
682 particularly with reference to CALL and RET pairing. (thanks to
683 Alastair Bridgewater)
684 * optimization: on x86, the code for access to thread-local symbol
685 values and binding/unbinding of thread-local symbols is smaller.
686 (thanks to Alastair Bridgewater)
687 * enhancement: CONSTANTP is now able to determine constantness of
688 more complex forms, including calls to constant-foldable standardized
689 functions and some special forms beyond QUOTE.
690 * fixed bug: occasional GC crashes on Solaris/x86
691 * optimization: x86-64 supports stack allocation of results of simple
692 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
693 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
694 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
695 (thanks to James Knight)
696 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
697 by the standard (thanks to Stephen Compall)
699 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
700 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
701 be used for bundling the runtime and the core file into one
702 executable binary. This feature is not currently supported on all SBCL
703 platforms. (thanks to James Bielman and NIIMI Satoshi)
704 * new feature: a generational or ephemeral garbage collector is now
705 the default on the PowerPC platform (both Linux and Darwin). The
706 old Cheney (stop and copy) collector is a build-time option.
707 (thanks to Cyrus Harmon, after Raymond Toy)
708 * minor incompatible change: the method by which SBCL finds its
709 contributed modules has changed; it no longer relies on symbolic
710 links from an $SBCL_HOME/systems directory, but searches directly
711 in the subdirectories of $SBCL_HOME.
712 * enhancement: the dynamic heap size on the Linux/PPC platform is
713 markedly larger, even using the older Cheney garbage collector.
714 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
715 certain complicated string types. (reported by Gary King)
716 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
717 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
718 * fixed bug: DOCUMENTATION on structures no longer signals an
719 error if no structure type of the right name exists. (reported by
721 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
722 applicable non-standard primary or :AROUND method on
723 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
724 no longer cause unbound variable errors. (reported by Kevin Reid)
725 * fixed bug: metacircle resolution in cases where methods have slots
726 added before the slots from STANDARD-METHOD. (reported by Jean
728 * fixed bug: the Power PC assembly code for calling into lisp
729 sometimes computed its return address wrongly (depending on the
730 15th bit of the address, fixed at link time). This used to
731 manifest itself in a segmentation violation while building PCL.
732 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
733 many others over the years)
734 * fixed bug: ROOM no longer reports silly numbers for stack usage if
735 the stack pointer (treated as a signed integer) is negative.
736 (thanks to Peter van Eynde)
738 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
739 * new platform: experimental support for the Windows operating
740 system has been added. (thanks to Alastair Bridgewater)
741 * fixed several bugs in and robustified the PPC FFI (including
742 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
743 * bug fix: allow non-simple string symbol names (reported by Paul
745 * bug fix: interrupt handling on NetBSD (thanks to Richard M
747 * bug fix: saving a core corrupted callbacks on x86/x86-64
748 * bug fix: closed a loophole in metacircularity detection and
749 grounding in the PCL implementation of CLOS.
750 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
751 (reported by tomppa on #lisp)
752 * bug fix: clear the direction flag on Lisp -> C transitions, as
753 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
754 (reported by Andrej Grozin and Hendrik Maryns)
755 * optimization: major improvements to GC efficiency on GENCGC platforms
756 * optimization: faster implementation of EQUAL
757 * optimization: emit more efficient opcodes for some common
758 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
759 * optimization: several other minor code-generation improvements on x86-64
760 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
762 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
763 * minor incompatible change: (SETF CLASS-NAME) and (SETF
764 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
765 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
766 by ANSI as a generic function, and for consistency (SETF
767 GENERIC-FUNCTION-NAME) is treated likewise.
768 * fixed bug #233.b: make constraint propagation notice when a variable
769 value is changed after it is referenced but before it is used
770 * fixed bug #296: no more arbitrary behaviour from filenames with
771 odd characters as --load arguments. (reported by Adam Warner)
772 * fixed bug #390: :CHARACTER-SET pathname components now work as
773 expected. (reported by Tim Daly Jr)
774 * fixed bug #391: complicated :TYPE intersections in slot
775 definitions no longer cause an error in PCL internals.
776 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
777 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
778 returning the number of octets which would be written to the
779 file-stream. (thanks to Robert J. Macomber)
780 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
781 arguments to the internal stream functions. (thanks to David
783 * bug fix: stack exhaustion is now handled on the x86/SunOS
784 platform. (thanks to Daisuke Homma)
785 * bug fix: more accurate ROOM results on GENCGC platforms
786 * optimization: improved type inference for arithmetic-for
787 index variables in LOOP
788 * optimization: faster floating-point SQRT on x86-64
789 * fixed some bugs revealed by Paul Dietz' test suite:
790 ** DOCUMENTATION returns NIL instead of "" for method combinations
791 that don't have a docstring
793 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
794 * minor incompatible change: (SETF CLASS-NAME) and (SETF
795 GENERIC-FUNCTION-NAME) are no longer generic functions, and
796 therefore are not customizeable by user code (as seems to be at
797 least permitted and maybe required by AMOP). As a consolation,
798 however, the SBCL implementation of these functions now calls
799 REINITIALIZE-INSTANCE as specified by AMOP.
800 * bug fix: slot-definition documentation is propagated as per ANSI
801 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
802 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
803 argument, as do eslotd initialization methods. (from Pascal
804 Costanza's "Closer" project)
805 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
806 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
808 * bug fix: it is now possible to have more than one subclass of
809 STANDARD-GENERIC-FUNCTION without causing stack overflow.
810 (reported by Bruno Haible, Pascal Costanza and others)
811 * bug fix: the dependent update protocol now works for generic
812 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
814 * bug fix: condition-class instances corresponding to
815 DEFINE-CONDITION forms are now created eagerly. (reported by
816 Kalle Olavi Niemitalo on comp.lang.lisp)
817 * bug fix: floating point printing is more accurate in some
818 circumstances. (thanks to Simon Alexander)
819 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
820 merged with *DEFAULT-PATHNAME-DEFAULTS*.
821 * bug fix: callbacks on OS X now preserve stack-alignment.
822 (thanks to Cyrus Harmon)
823 * enhancement: the x86-64 disassembler is much better at
824 disassembling SSE instructions. (thanks to Lutz Euler)
825 * enhancement: CHECK-TYPE on a variable now allows type-inference
826 (thanks to Vincent Arkesteijn)
827 * optimization: improved performance of EUC-JP external format.
828 (thanks to NIIMI Satoshi)
829 * optimization: performance improvements to IO on file streams of
830 :ELEMENT-TYPE CHARACTER
831 * optimization: much faster memory allocation on x86-64
832 * optimization: faster garbage collections (latency and throughput)
834 * optimization: faster bignum subtraction and division on x86 and x86-64
835 (thanks to Lutz Euler)
836 * optimization: more accurate type-inference for loops that use a
837 floating point index variable or a negative step.
839 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
840 * bug fix: add a workaround to SBCL looping infinitely at startup on
841 Linux kernels with apparently buggy implementations of personality().
842 (thanks to Svein Ove Aas)
843 * bug fix: Unicode symbols are correctly printed in LDB backtraces
844 (thanks to David Lichteblau)
845 * bug fix: local bindings shadow global symbol macros for
846 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
847 * optimization: non-open coded uses of numeric comparison operators
848 (e.g. >) no longer cons when called with more than one parameter
849 on platforms supporting dynamic-extent allocation.
850 * enhancement: saving cores with foreign code loaded is now supported
851 on MIPS/Linux in addition to the previously supported platforms.
852 * bug fix: FIND-RESTART now tests for activity, not applicability when given
853 a restart object as identifier. (reported by Helmut Eller for CMUCL)
854 * bug fix: division by zero in sb-sprof when no samples were collected
855 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
856 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
857 (thanks to Alexander Kjeldaas)
858 * new feature: ignoring the return values of destructive functions
859 like NREVERSE signals a compile-time style-warning.
860 (thanks to Kevin Reid)
861 * new feature: the alignment of alien structure fields can be
862 explicitly specified. (thanks to Cyrus Harmon)
863 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
864 and dump core on SIGQUIT
866 ** incompatible change: threads do not inherit values of specials
867 from their parents (see manual)
868 ** bug fix: threads stacks belonging to dead threads are freed by the
869 next exiting thread, no need to gc to collect thread stacks anymore
870 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
871 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
872 ** bug fix: dynamic variable and thread start related gc lossage
873 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
875 * fixed some bugs revealed by Paul Dietz' test suite:
876 ** SUBTYPEP is slightly more accurate on heinously complicated
877 CONS types where some of the members have uncertain (in the
878 NIL, NIL sense) type relationships to each other.
880 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
881 no longer purified unless :PURIFY T is explicitly specified.
882 ** Non-purified cores are significantly smaller than before
884 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
885 * new feature: timers based on Zach Beane's excellent timer package
886 * added support for the following external formats: koi8-u,
887 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
888 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
889 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
890 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
891 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
892 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
893 * incompatible change: a threaded SBCL will no longer revert to
894 non-threaded mode on non-NPTL systems, but refuse to start entirely.
895 * bug fix: interrupts are disabled until startup is complete; no
896 more sigsegvs when receiving a signal to soon
897 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
899 * bug fix: add a workaround for the memory randomization features in
900 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
901 workaround will only be in effect on systems with the proc filesystem
903 * bug fix: printing objects of type HASH-TABLE signals a
904 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
906 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
907 have been running for over 50 days. (reported by Gilbert Baumann)
908 * bug fix: the logic for getting names of functions gets less
909 confused when confronded with alternate-metaclass
910 funcallable-instances. (reported by Cyrus Harmon)
911 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
912 consistent, even on internal alternate-metaclass objects.
913 * bug fix: flush closure information collected by physical
914 environment analysis prepass before the main pass. (bug reported
916 * bug fix: compiler pack phase does not modify a hashtable, which is
917 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
918 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
919 may be expanded. (reported by Alexander <malishev> on c.l.l)
920 * bug fix: no more style-warnings when compiling code using LAST
921 under high-SPEED low-SPACE settings. (reported by David Wallin
923 * fixed some incompatibilities between SBCL's MOP and the MOP
925 ** the METAOBJECT class is now implemented;
926 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
927 STANDARD-OBJECT, as required;
928 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
929 now compatible (as required by VALIDATE-SUPERCLASS); there
930 remains a consistency requirement over the presence or absence
931 of the FUNCTION class in the superclass list of finalized
932 classes; see the manual for more details;
933 ** the :AROUND method for COMPUTE-SLOTS on
934 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
935 requested slot ordering.
937 ** bug fix: parent thread now can be gc'ed even with a live
939 ** bug fix: binding a special with PROGV to no value is not
940 the same as that symbol not having been bound (thanks to
942 ** bug fix: binding specials is thread safe (thanks to
944 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
945 and SB-SPROF do not die with 'no handler for signal XX in
946 interrupt_handle_now(..)' anymore
947 ** bug fix: WITH-TIMEOUT works with multiple running threads
948 * fixed some bugs revealed by Paul Dietz' test suite:
949 ** ENSURE-GENERIC-FUNCTION should take a method class object for
950 the :method-class keyword argument.
952 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
953 * new port: the Solaris operating system on x86 processors is now
954 mostly supported, though some rough edges in the environment
955 remain. (thanks to Daisuke Homma)
956 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
957 space for its heap. (thanks to Thiemo Seufer)
958 * optimization: SBCL on MIPS platforms now supports dynamic-extent
959 closures. (thanks to Thiemo Seufer)
960 * minor incompatible change: eof selects abort in the debugger.
961 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
962 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
964 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
965 deprecated, and will go away in a future revision of SBCL.
966 * minor incompatible change: GC-ON and GC-OFF are no longer
967 implemented with a counter, it does not matter how many times gc
968 is switched on or off
969 * bug fix: discriminating functions for generic function classes
970 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
971 make invalid assumptions about method precedence order. (reported
973 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
974 platforms. (thanks to Thiemo Seufer)
975 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
976 * bug fix: correct pathname printing with printer escaping is on.
977 (thanks to Kevin Reid)
978 * bug fix: complex VOP definitions in "user-space" no longer trigger
979 package locks. (reported by Zach Beane)
980 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
981 for funcallable-instance functions, and is no different from
983 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
984 characters in the current readtable. (reported by Nicholas Neuss)
985 * bug fix: SUBTYPEP on various CONS types returns more a more
986 accurate acknowledgment of its certainty.
987 * optimizations: REMOVE-DUPLICATES now runs in linear time on
988 lists in some cases. This partially fixes bug 384.
989 * flush all standard streams before prompting in the REPL and the
991 * bug fix: signal handling and triggering gc do not conflict
992 directly anymore, in particular a high frequency sb-sprof does
993 not prevent gc from running
994 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
995 approximation for timezone and DST information between the
996 universal time epoch and the smallest negative 32-bit time_t.
997 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
998 year 1899 so long as the encoded time is non-negative.
999 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1000 the correct answer for non-constant rotate amounts when performing
1001 an inline 32-bit rotation.
1003 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1004 there is only one thread in the session
1005 ** bug fix: memory leak for streams created in one thread and
1006 written to in another
1007 ** bug fix: lockup when compiled with gcc4
1008 ** bug fix: race that allows the gc to be triggered when gc is
1010 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1011 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1013 ** bug fix: finished threads are gc'ed properly
1014 * fixed some bugs revealed by Paul Dietz' test suite:
1015 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1016 has arguments with a different set of applicable methods from
1017 the orignal arguments.
1018 ** The type error thrown by MAP now has a correct expected-type
1020 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1021 name a compiled function.
1022 ** Three MISC tests where a large bignum was improperly coerced to
1023 a float that couldn't represent that bignum during type
1024 derivation were fixed.
1025 ** SUBTYPEP can now handle types involving the negation of a
1026 list-form FUNCTION type.
1027 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1028 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1029 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1031 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1032 * New feature: Experimental support for bivalent streams: streams
1033 opened with :element-type :default now allow character and binary
1034 (unsigned-byte 8) I/O
1035 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1036 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1038 * bug fix: no more highly sporadic "couldn't check whether ~S is
1039 readable" when reading a stream and an interrupt hits in the middle
1040 of a select system call
1041 * compiler better recognizes complex arrays (reported by Eduardo
1043 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1044 for variables on SBCL built with linkage-tables. (reported by Luis
1046 * various error reporting improvements.
1047 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1048 (Thanks to Thiemo Seufer)
1049 * bug fix: nested reader invokations work correctly
1050 * bug fix: it is possible to have simultaneous references to foreign
1051 code and foreign data with the same name.
1053 ** added x86-64 support
1054 ** incompatible change: the threading api now works with thread
1055 objects instead of thread ids
1056 ** bug fix: threads are protected from signals and interruption when
1057 starting up or going down
1058 ** bug fix: a race where an exiting thread could lose its stack to gc
1059 ** bug fix: don't halt on infinite error in threads if possible
1060 ** fixed numerous gc deadlocks introduced in the pthread merge
1061 ** bug fix: fixed thread safety issues in read and print
1062 ** bug fix: debugger doesn't hang on session lock if interrupted at
1063 an inappropriate moment
1064 ** bug fix: run-program is now thread safe(r)
1065 ** bug fix: inner with-recursive-lock no longer releases the mutex
1066 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1067 * fixed some bugs revealed by Paul Dietz' test suite:
1068 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1069 EXPECTED-TYPE slots filled.
1070 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1072 ** MAKE-CONDITION accepts classes as type-designators.
1073 ** COMPILE may never return NIL.
1074 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1075 range before calling Unix time functions
1077 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1078 * numerous signal handling fixes to increase stability
1079 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1080 * minor incompatible change: we now correctly canonize default
1081 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1082 as per the MOP, rather than the historical (INITARG INITFUNCTION
1083 INITFORM). (reported by Bruno Haible)
1084 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1085 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1086 for more information.
1087 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1088 pathname is a directory pathname.
1089 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1090 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1092 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1093 :METACLASS argument in addition to a class name. (reported by
1094 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1095 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1096 (reported by Bruno Haible; patch from Gerd Moellmann)
1097 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1099 * bug fix: more cleanups to the floating point exception handling on
1100 x86-64 (thanks to James Knight)
1101 * bug fix: the compiler does not try to constant fold calls of
1102 COERCE and BIT-* functions when they return freshly constructed
1103 objects. (reported by Nikodemus Siivola and Paul Dietz)
1104 * optimization: improved the MIPS versions of generic (in the
1105 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1106 * optimization: direct conversion of (unsigned-byte 32) to floats on
1107 the PowerPC platform.
1108 * optimization: structure instances with raw slots now use less
1109 memory, and probably show better memory locality. (thanks to
1111 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1112 are treated as requests for stack allocation on the x86-64,
1113 Alpha-32, PPC and SPARC platforms.
1114 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1115 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1116 (thanks to Tony Martinez)
1118 ** gcing a dead thread can no longer lead to lockups
1119 ** threads block signals until they are set up properly
1120 ** errno is no longer shared by threads
1121 ** interrupt-thread restores the eflags register on x86
1122 ** fixed some lockups due to gc/thread interaction
1123 * fixed some bugs revealed by Paul Dietz' test suite:
1124 ** invalid dotted lists no longer raise a read error when
1125 *READ-SUPPRESS* is T
1126 ** signal an error if a symbol that names a declaration is used
1127 as the name of a type, or vice versa
1128 ** allow using the (declare (typespec var*)) abbreviation for
1129 (declare (type typespec var*)) with all type specifiers
1130 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1131 ** always return NIL from PPRINT-POP when OBJECT is NIL
1132 ** don't signal errors when pretty-printing malformed LABELS,
1133 FLET or MACROLET forms
1134 ** declarations in a DOLIST body are also in scope for the
1136 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1138 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1141 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1142 * fixed cross-compiler leakages that prevented building a 32-bit
1143 target with a 64-bit host compiler.
1144 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1145 opened with :IF-EXISTS :OVERWRITE.
1146 * fixed bug 281: error for an invalid qualifier in a short-form method
1147 combination method is not signalled until the faulty method is called.
1148 * bug fix: iteration variable type inferrer failed to deal with open
1149 intervals. (reported by Alan Shields)
1150 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1151 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1152 or more runtime options were provided to the sbcl binary.
1153 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1154 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1155 * optimization: there's now a fast-path for fixnum arguments in the
1156 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1157 * optimization: the code generated on x86-64 is more compact thanks
1158 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1159 * bug fix: floating-point exception handling now partly works on
1160 x86-64. (Thanks to James Knight)
1161 * improvement to the MIPS backend from Thiemo Seufer: C-style
1162 64-bit long long arguments and return values to alien functions
1164 * fixed some bugs revealed by Paul Dietz' test suite:
1165 ** the type-error signalled from WARN has a filled-in DATUM slot.
1166 ** the type-error required when a stream is not associated with
1167 a file has the stream as its datum.
1168 ** type-errors on single-floats on x86-64 no longer have
1169 :INVALID-OBJECT as the datum
1170 ** the type-errors signalled for invalid function names now have
1171 a correct expected type
1172 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1173 for structures defined with a :TYPE. Documentation strings for
1174 typed structures are no longer immediately discarded
1175 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1176 broadcast streams and synonym streams. FILE-LENGTH now also works
1177 on broadcast streams.
1179 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1180 * incompatible change: the --noprogrammer option, deprecated since
1181 version 0.7.5, has been removed. Please use the equivalent
1182 --disable-debugger option instead.
1183 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1185 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1186 inoperational for a while now) has been completely removed.
1187 * null lexical environments are now printed as #<NULL-LEXENV>,
1188 significantly reducing the amount of clutter in typical backtraces.
1189 * documentation on weak pointers, finalization, and after GC hooks
1190 has been added to the manual.
1191 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1192 as other specialized array types, is much faster. SUBSEQ and
1193 COPY-SEQ on such arrays have also been sped up.
1194 * optimization: EQL is now more efficient when at least other argument
1195 is known to be of type (OR FIXNUM (NOT NUMBER)).
1196 * fixed bug: the runtime is now less vulnerable to changes in the
1197 size of the SBCL object on OS X, and virtual memory is reserved for
1198 all spaces that need to be at a fixed address.
1199 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1200 * fixed bug: finalizers and after GC hooks that cause consing are now
1202 * fixed bug: compiler error messages and summaries are now printed to
1203 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1204 * fixed inference of the upper bound of an iteration variable.
1205 (reported by Rajat Datta).
1206 * fixed bug 211e: calling local functions with duplicated constant
1207 keyword argument no longer causes a bogus style warning about an
1209 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1210 ftype declaration to be disregarded. (reported by Dave Roberts)
1211 * fixed bug 373: caused by erronous compilation of references to alien
1212 variables in the runtime on ppc/darwin.
1213 * fixed bug 376: CONJUGATE type deriver.
1214 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1215 (reported by Baughn on #lisp)
1216 * a cross-compiler bug on non-x86 platforms has been identified and
1217 fixed. (thanks to Bruno Haible)
1218 * improvements to the MIPS runtime code for increased stability.
1219 (thanks to Thiemo Seufer)
1220 * increased the maximimum compact environment size to allow
1221 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1222 * partial workaround for bug 135: don't name the function we're
1223 calling for hairy cases of EVAL, so as not to accumulate one environment
1224 entry per call to EVAL. (reported by Kevin Reid)
1225 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1226 * optimization: type testing for non-vector arrays should be faster.
1227 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1229 * bug fix: setting 31st element of a bit vector to zero did not work
1231 * bug fix: redefining a class definition which failed due to a
1232 previous accessor / function clash now works (but see BUGS entry
1233 #380 for more problems in this area). (thanks to Zach Beane)
1234 * the long form of DEFINE-METHOD-COMBINATION disables method group
1235 checking when given a single method group with pattern *. (thanks
1237 * on x86 compiler supports stack allocation of results of simple
1238 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1239 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1240 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1241 * fixed some bugs related to Unicode integration:
1242 ** the restarts for recovering from input and output encoding
1243 errors only appear when there is in fact such an error to
1245 * fixed some bugs revealed by Paul Dietz' test suite:
1246 ** MISC.549 and similar: late transformation of unsafe type
1247 assertions into derived types caused unexpected code
1249 ** SCALE-FLOAT type deriver is less wrong.
1250 ** type derivers for EXP, LOG and similar functions compute result
1251 types for complex arguments better.
1252 ** (MISC.563) CONJUGATE type deriver works for very restricted
1254 ** out-of-line type testers for character strings are available.
1255 ** EQUAL compiler transform understands specialness of objects
1257 ** accessing double-floats stored on the stack now works on x86-64.
1258 ** debugger internals could sometimes create invalid lispobjs,
1259 resulting in GC crashes.
1260 ** MISC.548: type check weakening can convert required type into
1262 ** initialization forms for bindings are not in scope of free special
1265 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1266 * incompatible change: thread support for non-NPTL systems has
1267 been removed - locking is buggy and unreliable. A threaded
1268 SBCL build will now warn at startup and refuse to create
1269 new threads, unless futex support is detected in the kernel
1270 * incompatible change: the top level REPL now has only an ABORT
1271 restart associated with it, not TOPLEVEL and ABORT as it used to.
1272 TOP and TOPLEVEL are now available as debugger commands for
1273 returning to the top level.
1274 * incompatible change: forms evaluated in the REPL now use the
1275 global optimization policy.
1276 * incompatible change: user- and system-initialization files are
1277 no longer processed with LOAD, but by READ and EVAL; hence the
1278 global optimization policy, startup package, readtable, etc,
1280 * The .fasl file format number has been incremented because of
1281 various incompatible changes.
1282 * internal entry point details and argument counts no longer appear
1283 in backtraces unless explicitly requested by setting
1284 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1285 * built-in and standard functions no longer have names like "top
1286 level local call to FOO".
1287 * fixed bug 32: functions defined in non-null lexical environments
1288 now have more legible printed representation
1289 * fixed bug 33: functions defined in non-null lexical environemnts
1290 are now more amenable to inspection by INSPECT.
1291 * workaround for bug 354: XEPs no longer appear in backtraces unless
1292 explicitly requested.
1293 * fixed bug: receiving the signal which results from attempting to
1294 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1295 SIGBUS on other *BSDs) on architectures where the C stack is also
1296 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1297 to debugger confusion or wild instability. (reported by Cheuksan
1299 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1300 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1301 (reported by Lutz Euler)
1302 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1303 compiled code. (reported by Paul Dietz)
1304 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1305 specialized parameter to a method to something that is not TYPEP
1306 the specializer is now possible.
1307 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1308 face of package deletion.
1309 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1310 pathname. (reported by David Tolpin on c.l.l)
1311 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1312 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1313 STRING-STREAM and FILE-STREAM, respectively.
1314 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1315 than previously; in addition, its test suite is now run on build.
1316 (thanks to Luke Gorrie)
1317 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1318 * optimization: added a immediate representation for single-floats
1320 * fixed some bugs revealed by Paul Dietz' test suite:
1321 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1322 ** attempting to create a package with a colliding nickname causes
1323 correctable errors to be signalled.
1324 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1325 ** several x86-64 backend bugs related to sign-extension of immediate
1328 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1329 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1330 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1332 * fixed loading of multiply forward-referenced layouts.
1333 (thanks to Cheuksan Wang)
1334 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1335 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1336 (see COMPILE-FILE documentation for details of :PRINT option)
1337 * fixed bugs 19 and 317: fixed-format floating point printing is
1338 more accurate. This also fixes a bug reported by Adam Warner
1339 related to the ~@F format directive.
1340 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1342 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1343 dispatch table if the from-char is a dispatch macro character.
1344 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1345 by the wordsize no longer ignore the last word. (reported by Lutz
1347 * fixed bug in type checking of dynamic-extent variables. (reported
1349 * optimization: sequence traversal functions use their freedom to
1350 coerce function designators to functions.
1351 * optimization: code with many calls to CLOS methods specialized on
1352 CLOS classes has had redundant type checks removed.
1353 * optimization: added declarations to speed up operations that access
1354 the internal character database (for example STRING-UPCASE)
1355 * optimization: comparison operations between floats and sufficiently small
1356 fixnums no longer create extra rationals
1357 * fixed some bugs related to Unicode integration:
1358 ** portions of multibyte characters at the end of buffers for
1359 character-based file input are correctly transferred to the
1360 start of the buffer at the next read.
1361 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1362 passing it through to OPEN.
1363 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1364 argument given to internal calls to OPEN.
1365 ** the built-in comment readers (introduced by character sequences
1366 ";" and "#|") are more forgiving to encoding errors; they will
1367 STYLE-WARN and then attempt to resync the stream at a character
1368 boundary. (thanks to Teemu Kalvas)
1369 * fixed some bugs in the x86-64 port:
1370 ** Negative short int return values from c-calls are sign-extended
1372 ** The stack is aligned to 16-bytes for c-calls, as required by
1373 the ABI. (thanks to Cheuksan Wang)
1374 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1375 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1376 for enums. (thanks to Vincent Arkesteijn)
1377 ** Multiple small optimizations and bugfixes for floating point
1379 * fixed some bugs revealed by Paul Dietz' test suite:
1380 ** Space, Tab, Linefeed, Return and Page have the invalid
1381 secondary constituent character trait.
1382 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1384 ** WITH-INPUT-FROM-STRING should only update the index place on
1386 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1387 works more reliably.
1388 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1389 with one-dimensional arrays with fill pointers.
1390 ** TYPE-OF failed on a complex with an integer realpart and a
1392 ** compiler failure during type inference for the code of form
1393 (IF (EQL X (THE ...)) ...) (MISC.535).
1395 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1396 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1397 platform. The port supports 61-bit fixnums, large memory spaces
1398 and reloading shared object files.
1399 * enhancement: saving cores with foreign code loaded is now
1400 supported on ppc/Darwin in addition to the previously supported
1402 * enhancement: the statistical profiler now walks deeper into the
1403 call stack for more meaningful call-graphs and accrued time
1404 reports (x86/x86-64 only). It also now reports time spent in
1406 * enhancement: it is now possible to trace most individual methods
1407 of a generic function in addition to tracing the generic function
1409 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1411 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1412 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1413 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1414 produces an error. (thanks to Vincent Arkesteijn)
1415 * bug fix: NAMESTRING on pathnames with :WILD components in their
1416 directories works correctly. (thanks to Artem V. Andreev)
1417 * fixed bug 125: compiler preserves identity of closures. (reported
1419 * bug fixed: functions with &REST arguments sometimes failed with
1420 "Undefined function" when compiled with (DEBUG 3). (reported by
1422 * bug fix: overflow during compiling of setting element of a bit
1423 vector with constant index and value. (reported by Timmy Douglas)
1424 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1425 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1426 * on x86 compiler partially supports stack allocation of dynamic-extent
1428 * GO and RETURN-FROM do not check the extent of their exit points
1429 when compiled with SAFETY 0.
1430 * fixed some bugs related to Unicode integration:
1431 ** encoding and decoding errors are now much more robustly
1432 handled; it should now be possible to recover even from invalid
1433 input or output to the terminal. (thanks to Teemu Kalvas)
1434 ** provided a first cut at implementing STRING-TO-OCTETS and
1435 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1436 ** altered the SB-MD5 contributed module slightly, changing the
1437 interface just enough for it to be supportable for builds where
1438 lisp characters are not eight bits.
1439 * fixed some bugs revealed by Paul Dietz' test suite:
1440 ** the FORMATTER-generated functions for ~V[ conditionals require
1441 the correct number of arguments.
1442 ** READ-FROM-STRING returns the mandated second value when applied
1443 to displaced strings.
1444 ** the #\Rubout and #\Backspace characters are treated as invalid
1445 constituent characters by the tokenizer.
1447 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1448 * new feature: reloading changed shared object files with
1449 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1450 * new feature: references to foreign variables and functions
1451 can now be compiled and loaded before the corresponding shared
1452 object file is loaded, as long as the foreign definitions are
1453 available at runtime.
1454 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1455 supported. (thanks to Dan Debertin)
1456 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1457 just keywords. (thanks to Vincent Arkesteijn)
1458 * fixed bug #331: structure-class instances corresponding to
1459 DEFSTRUCT forms are now created eagerly.
1460 * fixed bug #345: backtraces from calls to undefined functions work
1461 on x86 as well. Related bug #61 is now also partially fixed on x86
1462 (backtraces from throws to unknown catch tags.)
1463 * bug fix: lambda-list parsing is now stricter vrt. order and number
1464 of lambda-list keywords.
1465 * bug fix: as specified by AMOP, an error is signalled if a
1466 class-option appears multiple times in a DEFCLASS form. (reported
1468 * bug fix: RANDOM can be compiled when the compiler derives the type
1469 of its numeric argument as a disjoint set of small integers.
1470 (reported by Paul Dietz)
1471 * bug fix: starting a core saved with shared objects loaded when
1472 those objects are not available no longer causes threaded SBCL to
1473 hang. (reported by Sean Ross)
1474 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1476 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1477 argument, which must always be NIL. (reported by Kalle Niemitalo)
1478 * bug fix: printing 1.0d+23 no longer results in an error.
1479 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1480 * bug fix: structure slot setters preserve evaluation order. (thanks
1482 * fixed some bugs related to Unicode integration:
1483 ** RUN-PROGRAM can allow its child to take input from a Lisp
1484 stream. (reported by Stefan Scholl)
1485 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1486 locales. (reported by Ken Causey)
1487 * fixed some bugs revealed by Paul Dietz' test suite:
1488 ** INCF, DECF and REMF evaluate their place form as specified in
1490 ** FORMATTER expands ~{ ~} iteration directives with V or #
1491 parameters correctly.
1492 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1493 consequent uses no arguments correctly.
1494 ** the system has a partial understanding of the (COMPLEX RATIO)
1496 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1499 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1500 * new feature: a build-time option (controlled by the :SB-UNICODE
1501 keyword feature, enabled by default) for building the system with
1502 support for the entire 21-bit character space defined by the
1504 * new feature: the system now has rudimentary external-format
1505 support; the primary user-visible change at this time is that
1506 characters with the high bit set (such as lower-case-e-acute) will
1507 print correctly to a terminal in a UTF-8 environment.
1508 * minor incompatible change: BASE-CHAR no longer names a class;
1509 however, CHARACTER continues to do so, as required by ANSI.
1510 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1511 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1512 has been moved to the SB-EXT package (temporarily re-exported from
1514 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1515 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1517 * fixed bug #351: better error-handlind and reporting for malformed
1519 * fixed bug #350: bignum-printing is now more memory-efficient,
1520 allowing printing of very large bignums, eg. (expt 2 10000000).
1521 (reported by Bruno Haible)
1522 * fixed bug #302: better primitive-type selection for intersection
1524 * fixed bug #308: non-graphic characters now all have names, as
1525 required. (reported by Bruno Haible)
1526 * bug fix: redefining a class with different superclasses now correctly
1527 removes it from the direct-subclasses of its previous superclasses.
1528 (reported by David Morse)
1529 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1530 new value now works. (reported by Bruno Haible)
1531 * bug fix: correct canonicalization of multiple non-standard slot
1532 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1533 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1534 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1535 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1536 now exists, an signals an error.
1537 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1538 during macroexpansion for non-iteration variables as well. (reported
1539 by Bruno Haible for CMUCL)
1540 * bug fix: Cyclic structures and unprintable objects in compiler
1541 messages no longer cause errors. (reported by Bruno Haible)
1542 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1543 and READ-FROM-STRING all now return a primary value of NIL if
1544 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1545 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1546 (reported by Bruno Haible for CMUCL)
1547 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1548 do not designate a condition. (reported by Bruno Haible for
1550 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1551 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1552 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1553 permitting resolution in favour of any of the conflicting symbols.
1554 (reported by Bruno Haible for CMUCL)
1555 * bug fix: EQUAL compiler optimizations is less aggressive on
1556 strings which can potentially compare true despite having distinct
1557 specialized array element types.
1558 * bug fix: unit enumerations can be defined without dividing by
1559 zero. (thanks to Vincent Arkesteijn)
1560 * FORMAT compile-time argument count checking has been enhanced.
1561 (report from Bruno Haible for CMUCL)
1562 * a partial workaround for the bug 262: the compiler does not try to
1563 inline-expand a local function doing RETURN-FROM from a deleted
1564 BLOCK. (thanks to Peter Denno for the bug report and to David
1565 Wragg for the simple test case)
1566 * fixed some bugs revealed by Paul Dietz' test suite:
1567 ** RENAME-PACKAGE allows all package designators as new package
1569 ** constraint propagation and lambda variable substitution are
1570 more cautious in dealing with partially deleted code.
1571 ** compiler sometimes lost reoptimization passes.
1572 ** CERROR, given a condition as condition designator, passes its
1573 remaining arguments to the continue format control without
1575 ** Case-altering FORMAT directives work correctly on non-ASCII
1577 ** The REAL type specifier handles bounds outside the floating
1578 point ranges without signalling FLOATING-POINT-OVERFLOW.
1579 ** Functions with IR1-transformations can create intercomponent
1580 references to global functions.
1581 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1583 ** FORMAT ~R treats a nil value for its first parameter correctly.
1585 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1586 * enhancement: saving cores with foreign code loaded is now
1587 supported on x86/NetBSD and sparc/Linux in addition to the previously
1588 supported platforms.
1589 * bug fix: on some platforms repeated installations caused multiple
1590 copies of HTML documentation to be installed -- should not happen
1591 any more. (reported by Stefan Scholl)
1592 * bug fix: parsing self-recursive alien record types multiple times
1593 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1594 original patch by Helmut Eller for CMUCL)
1595 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1596 (thanks to Richard Kreuter)
1597 * bug fix: defining classes whose accessors are methods on existing
1598 generic functions in other (locked) packages no longer signals
1599 bogus package lock violations. (reported by François-René Rideau)
1600 * bug fix: special variables as DEFMETHOD parameters no longer have
1601 associated bogus type declarations. (reported by David Wragg and
1603 * bug fix: read-write consistency on streams of element-type
1604 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1605 * bug fix: redefiniton of the only method of a generic function with
1606 no DEFGENERIC no longer emits a full WARNING. In addition,
1607 redefinition of generic functions with no DEFGENERIC to an
1608 incompatible lambda list now signals an error. (thanks to Zach
1610 * bug fix: DEFGENERIC now works even when there's a function of the
1611 same name in an enclosing lexical environment. (thanks to Zach
1613 * fixed compiler failure, caused by instrumenting code during
1614 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1615 * optimization: added loop analysis and improved register allocation
1616 * fixed some bugs revealed by Paul Dietz' test suite:
1617 ** POSITION on displaced vectors with non-zero displacement
1618 returns the right answer.
1619 ** (SIMPLE-STRING) is a valid type specifier for sequence
1621 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1623 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1624 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1626 ** The justification version of the FORMAT ~< directive treats
1627 non-zero minpad parameter correctly.
1629 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1630 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1631 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1632 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1633 the supported interface.
1634 * new feature: Single-stepping of code compiled with DEBUG 2 or
1635 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1636 * new feature: saving cores with foreign code loaded is now
1637 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1638 Timothy Moore's work for CMUCL)
1639 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1640 parameters to * instead of NIL if no initform is supplied.
1641 (reported by Johan Bockgaard)
1642 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1643 correctly when FUNCALL appears as the car of the form. Note:
1644 despite this FUNCALL forms are not currently subject to
1645 compiler-macro expansion. (port of Raymond Toy's fix for the
1646 same from CMUCL, reported by Johan Bockgaard)
1647 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1648 (thanks for Teemu Kalvas)
1649 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1650 case instead of terminating the iteration (thanks for Julian
1651 Squires, Sean Champ and Raymond Toy)
1652 * bug fix: incorrect expansion of defgeneric that caused a style
1653 warning. (thanks for Zach Beane)
1654 * bug fix: slot accessor effective method computation works properly
1655 for classes with multiple non-standard applicable methods on
1656 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1657 * on x86 compiler supports stack allocation of results of LIST and
1658 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1659 CMUCL implementation by Gerd Moellmann)
1660 * fixed some bugs revealed by Paul Dietz' test suite:
1661 ** FORMAT strings with both the ~<~:;~> form of the justification
1662 directive and pretty-printing directives cause an error.
1664 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1665 * incompatible change: the internal functions
1666 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1667 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1668 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1669 instead of the old functions.
1670 * new feature: on platforms where "dladdr" is available foreign
1671 function names now appear in backtraces. (based on Helmut Eller's
1673 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1674 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1676 * documentation: ASDF manual is now distributed with SBCL.
1677 * bug fix: SBCL can now load its contributed modules with REQUIRE
1678 even if the system-provided entries have been removed from
1679 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1681 * bug fix: backtraces involving undefined functions or assembly
1682 routines are more informative. (thanks to Brian Downing)
1683 * bug fix: mutually referent alien structures now work correctly.
1684 (reported by Rick Taube)
1685 * bug fix: structures defined by WITH-ALIEN can be referred to
1686 within other definitions in the same WITH-ALIEN.
1687 * bug fix: division operators (MOD, TRUNCATE and the like) with
1688 constant zero divisors and integer dividends no longer generate
1690 * bug fix: provide default methods for INPUT-STREAM-P and
1691 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1692 * bug fix: improve the detection and resolution of MOP metacycles
1693 (where effective method computation depends on the generic function
1694 having its effective method computed). (reported by Bruno Haible)
1695 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1696 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1697 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1698 * bug fix: pathnames with relative directory components can be
1699 represented relative to default pathnames.
1700 * optimization: in taking the GCD of bignums, reduce the two bignums
1701 to approximately the same size (using Euclid's algorithm) before
1702 applying the more sophisticated binary GCD. (thanks to Juho
1704 * optimization: COUNT on bitvectors now operates word-at-a-time.
1705 * optimization: ASH with a positive, but not necessarily constant,
1706 (leftwards) shift, when in a modular context, is compiled to a
1708 * fixed some bugs revealed by Paul Dietz' test suite:
1709 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1710 if the corresponding argument is NIL.
1711 ** FORMAT directives accepting a minpad parameter treat negative
1713 ** literal commas as character parameters to FORMAT directives are
1715 ** literal spaces directly after ~<Newline> directives within a
1716 format-logical-block (~:< ~@:>) do not induce :FILL-style
1717 conditional newlines.
1718 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1719 as specified: it no longer includes conditional newlines).
1720 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1722 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1723 :PER-LINE-PREFIX argument does not evaluate to a string.
1725 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1726 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1727 the manual for details; for now, package locks can be disabled by
1728 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1729 if no major problems are found then it is likely that they will be
1730 compiled in unconditionally.
1731 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1732 unsupported operators on all platforms. To load a shared library
1733 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1734 object file, link it into a shared library outside of SBCL and
1735 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1737 * bug workaround: ROOM T and the GC were not working together
1738 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1739 reporting facility aren't preserved. That reporting has been
1740 disabled, so now until and unless someone figures out how to make it
1741 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1742 * minor incompatible change: as threatened around sbcl-0.8.0, the
1743 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1744 The symbols are also exported from SB-PCL for backwards
1745 compatibility, but more so than before SB-PCL should be treated as
1746 an implementation-internal package.
1747 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1749 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1750 It is known as of this release not to work on the Alpha, however.
1751 * fixed bug #167: errors signalled due to illegal syntax in method
1752 bodies are now more legible.
1753 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1754 designators and can hence be used with TYPEP.
1755 * fixed bug #333: CHECK-TYPE now ensures that the type error
1756 signalled, if any, has the right object to be accessed by
1757 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1758 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1759 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1761 * fixed bug #334: programmatic addition of slots using specialized
1762 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1763 and :ALLOCATION :CLASS slots.
1764 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1765 float exponents. (rereported by Peter Seibel)
1766 * fixed a bug: #\Space (and other whitespace characters) are no
1767 longer considered to be macro characters in standard syntax by
1768 GET-MACRO-CHARACTER.
1769 * fixed bug: initialization of condition class metaobjects no longer
1770 causes an instance of the condition to be created. (reported by
1772 * fixed bug: it is now possible to have slots such that
1773 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1774 description is neither :INSTANCE nor :CLASS.
1775 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1776 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1777 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1778 non-standard SLOT-DEFINITION-ALLOCATION values.
1779 * fixed bugs: various race conditions handling exiting threads.
1780 CL-PPCRE's thread torture test now passes.
1781 * fixed arguably-a-bug: GC time proportional to number of threads in
1782 system even when most of them are idle
1783 * optimization: improved performance of BIT and SBIT on bit-vectors.
1784 * .fasl file incompatibility: The fasl file version number has
1785 been incremented because of changes associated with package locks.
1787 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1788 * minor incompatible change: the system no longer provides
1789 optimization hints (conditions which are TYPEP
1790 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1791 mode; these hints are still emitted when higher SPEED optimization
1793 * new contrib module: a sampling profiler (profiling by statistical
1794 sampling, rather than by instrumenting functions) is available as
1795 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1796 * the behaviour of the standard function ED is now customizeable by
1797 third parties through a hook variable: see ED's documentation
1798 string for information on the protocol.
1799 * the compiler no longer emits efficiency notes for (FUNCALL X)
1800 when the type of X is uncertain under default optimization
1802 * fixed bug 276: mutating a binding of a specialized parameter to a
1803 method to something that is not TYPEP the specializer is now
1805 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1806 exhibit the required behaviour on the x86 platform. (thanks to
1807 Peter van Eynde, Eric Marsden and Bruno Haible)
1808 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1809 even for difficult arguments. (reported by Peter Graves)
1810 * fixed bug 141a: the backquote printer now descends quoted
1812 * fixed another bug in backquote printing: no more destructive
1813 modification of the form's list structure. (reported by Brian
1815 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1816 move between its address being taken and the call to
1817 interrupt_thread, fixing a crashing race condition.
1818 * the SB-POSIX contrib implementation has been adjusted so that it
1819 no longer exhibits ridiculously poor performance when constructing
1820 instances corresponding to C structs.
1822 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1823 * minor incompatible change: the sb-grovel contrib now treats C
1824 structures as alien (in the sense of SB-ALIEN) objects rather than
1825 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1826 has implications for memory management of client code
1827 (sb-grovel-returned objects must now be manually managed) and for
1828 type safety (alien objects now have full types).
1829 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1830 used to control emission of compiler diagnostics, rather than the
1831 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1832 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1833 quality should be considered deprecated.
1834 * install.sh now installs the user manual as well
1835 * (not quite a new documentable feature, but worth considering in
1836 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1837 beginnings of a semantically meaningful condition hierarchy is
1838 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1840 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1841 designator as the defaults argument.
1842 * fixed bug: Displaced arrays whose displaced-to array has become
1843 too small now cause ARRAY-DIMENSION to signal an error, providing
1844 for safer bounds-checking. (reported by Bruno Haible)
1845 * fixed bug: DEFCLASS slot definitions with identical :READER and
1846 :WRITER names now signal a reasonable error. (reported by Thomas
1848 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1850 * fixed bug: Invalid slot specification errors now print correctly.
1851 (thanks to Zach Beane)
1852 * fixed bug 320: Shared to local slot value transfers in class
1853 redefinitions now happen corrently from superclasses as
1854 well. (reported by Bruno Haible)
1855 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1857 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1858 improper lists correctly. (reported by Bruno Haible)
1859 * fixed bug 313: source-transform for <fun-name> was erroneously
1860 applied to a call of a value of a variable with name <fun-name>.
1861 (reported by Antonio Menezes Leitao)
1862 * fixed bug 307: The obsolete instance protocol ensures that
1863 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1864 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1865 remove dead unknown-values globs from the middle of the stack.
1866 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1867 easier to resume long computations after using *BREAK-ON-SIGNALS*
1868 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1869 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1870 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1871 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1872 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1873 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1875 * fixed a (fairly theoretical) bug in string printing: if
1876 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1877 does not have array-element-type equal to the most general string
1879 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1880 function name. (thanks to Nikodemus Siivola)
1881 * fixed bug: compile-time format string checker failed on
1882 non-closed ~{. (reported by Thomas F Burdick)
1883 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1884 don't warn on legal loop code involving more than one aggregate
1885 boolean. (thanks to Nikodemus Siivola)
1886 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1887 in keyword position to MAKE-PACKAGE was not being reported as
1889 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1890 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1892 * fixed bugs 280 and 312: the checking for multiple definitions in a
1893 file is less likely to become confused by uses of inline
1895 * fixed bug: the #S reader macro performs the keyword coercion
1896 specified for slot names. (reported by Kalle Niemitalo)
1897 * fixed bug: lambda lists may contain symbols whose names start with
1898 & but are not lambda-list-keywords; their occurrence triggers a
1900 * fixed bug 321: define-method-combination argument lambda lists do
1901 not cause invalid code to be generated when &optional or &aux
1902 variables have default values. (reported by Bruno Haible)
1903 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1904 have CLOS classes; this appears also to have fixed reported
1905 instability in stack exhaustion detection.
1906 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1907 MUFFLE-WARNING when no associated restart is present is now
1909 * optimization: rearranged the expansion of various defining macros
1910 so that each expands into only one top-level form in a
1911 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1913 * optimization: used a previously little-used slot in symbols to
1914 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1916 * fixed some bugs revealed by Paul Dietz' test suite:
1917 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1919 ** Fixed an optimization bug related to inheritance of initforms
1920 from local to shared slots.
1921 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1922 ** CAST optimizer forgot to flush argument derived type.
1923 ** print/read consistency on floats is now orders of magnitude
1924 more likely. (thanks also to Bruno Haible for a similar report
1926 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1927 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1928 for objects of type REAL. Make it so.
1929 ** SXHASH is specified (infelicitously) to respect similarity,
1930 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1931 it so. (thanks to Markus Ziegler)
1932 ** on the Alpha, the compiler succeeds in compiling functions
1933 returning a known number of arguments greater than 63.
1934 ** fixed handling of invalid NIL arguments in keyword position
1936 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1938 ** provided a readably-printable representation for RANDOM-STATE
1940 ** ensured that pathnames, where they have a namestring, always
1941 print using #P"..." syntax.
1943 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1944 * Support for the forthcoming 2.0 version of the NetBSD kernel
1945 running on x86 hardware has been added. (thanks to Perry
1946 E. Metzger most immediately, and others for their past work)
1947 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1948 current release version; SBCL's OpenBSD support had been broken
1949 since about the time of OpenBSD's switch to ELF binary format.)
1950 * [placeholder for DX summary]
1951 ** user code with &REST lists declared dynamic-extent, under high
1952 speed or space and low safety and debug optimization policy.
1953 * The manual has been converted to Texinfo format and the debugger
1954 chapter from the cmucl manual has been added.
1955 * A facility has been added to extract documentation strings from
1956 sbcl and store them as Texinfo-formatted snippets for inclusion in
1957 the manual (via Texinfo's @include directive)
1958 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1959 the test case to Dave Roberts)
1960 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1961 pointers. (reported by Sean Ross)
1962 * bug fix: PROFILE output is printed nicely even for large numerical
1963 values. (thanks to Zach Beane)
1964 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1965 greater than 32 handle EOF correctly.
1966 * bug fix: on X86 an immediate argument of the IMUL instruction is
1967 correctly printed in disassembly. (thanks to Lutz Euler)
1968 * bug fix: class slots in redefined classes preserve their old
1969 values. (thanks to Bruno Haible and Nikodemus Siivola)
1970 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1971 succeeds. (reported by Marco Baringer)
1972 * fixed some bugs revealed by Paul Dietz' test suite:
1973 ** READ-SEQUENCE now works on ECHO-STREAMs.
1974 ** RATIONALIZE works more according to its specification. (thanks
1977 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1978 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1979 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1980 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1981 should matter to you only if you rebind the printer control
1982 variables and then find you want different bindings in the
1983 debugger than in the ordinary execution of your program.)
1984 * The runtime build system has been tweaked to support building
1985 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1986 assembler and linker. (thanks to Nikodemus Siivola)
1987 * Unbound, undefined, undeclared variables now trigger full
1988 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1989 more often programmer error than deliberate exploitation of undefined
1991 * optimization: the hash algorithm for strings has changed to one
1992 that is less vulnerable to spurious collisions. (thanks to Juho
1994 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1995 needless bounds checking. (thanks to Juho Snellman)
1996 * optimization: implemented multiplication as a modular
1997 (UNSIGNED-BYTE 32) operation on the PPC backend.
1998 * fixed some bugs revealed by Paul Dietz' test suite:
1999 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2001 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2002 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2003 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2005 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2007 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2009 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2010 their output stream on EOF from read.
2011 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2012 have been read to end-of-file.
2013 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2015 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2016 description of determination of which consecutive characters
2018 ** Printing the "Space" character with escaping on now yields "#\\ ",
2019 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2020 ** Reading floating-point numbers with *READ-BASE* set to a number
2021 less than 10 works correctly.
2022 ** Reading floating-point numbers with *READ-BASE* set to a number
2023 more than 10 works correctly.
2024 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2025 the readtable currently in effect.
2027 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2028 * minor incompatible change: parsing of namestrings on a physical
2029 (Unix) host has changed; numbers after the final #\. in a
2030 namestring are no longer interpreted as a version field. This is
2031 intented to be largely invisible to the user, except that the
2032 meaning of the namestring "*.*.*" has changed: it now refers to a
2033 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2034 should usually be replaced by
2035 (make-pathname :name :wild :type :wild :version :wild)
2036 with the added benefit that this is more likely to be portable.
2037 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2038 option to OPEN now signals an error if the file being opened
2039 exists; this may have an impact on existing code.
2040 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2041 signals received, on the PowerPC platforms (both Linux and
2042 Darwin). (thanks to Pierre Mai for pointing out the location of
2044 * several fixes on OS X: The system now builds and runs cleanly on
2045 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2046 (thanks to Brian Mastenbrook)
2047 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2048 second-resolution: integer multiples of 1/3600 between -24 and 24.
2049 (thanks to Vincent Arkesteijn)
2050 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2051 type when called with 1 argument; PEEK-CHAR checked type of
2052 PEEK-TYPE only after having read first character from a
2053 stream. (reported by Peter Graves)
2054 * bug fix: the garbage collector now has much better locality
2055 behaviour, and in particular no longer treats all memory as being
2056 exhausted when many small objects point to each other in a deeply
2058 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2059 recognized as being TYPEP their class.
2060 * bug fix: the PUSHNEW documentation string has been corrected.
2061 (thanks to Vincent Arkesteijn)
2062 * bug fix: defaulting of the value for the last of an atypically
2063 large number of multiple values being bound was not being
2064 performed correctly on the Alpha or PPC platforms
2065 * optimization: implemented multiplication as a modular
2066 (UNSIGNED-BYTE 32) operation on the x86 backend.
2067 * optimization: SEARCH on simple-base-strings can now be open-coded.
2068 (see also contrib/compiler-extras.lisp for inspiration for
2069 teaching the compiler about the Boyer-Moore algorithm).
2070 * value, returned by MAX (and MIN) called with several EQUALP, but
2071 not EQL, arguments now does not depend on compiler settings.
2072 * fixed some bugs revealed by Paul Dietz' test suite:
2073 ** in stack analysis liveness information is propagated from
2074 non-local entry points.
2075 ** pathwise CAST removing failed when the CAST node did not start
2077 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2078 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2080 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2081 host is already defined.
2082 ** RENAME-FILE works on streams instead of signalling an internal
2084 ** PEEK-CHAR uses the current readtable when determining whether
2085 or not a character is whitespace.
2086 ** MERGE-PATHNAMES handles the case when the pathname does not
2087 specify a name while the default-pathname specifies a version.
2088 ** Pathnames now stand a better chance of respecting print/read
2090 ** Attempting to use standardized file system operators with a
2091 pathname with invalid :DIRECTORY components signals a
2093 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2095 ** DIRECTORY on logical pathnames is more correct.
2096 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2097 signal a TYPE-ERROR if their argument is not a stream.
2098 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2099 designator argument does not designate a stream.
2100 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2101 examining the synonym.
2102 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2104 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2105 element-type larger than ([UN]SIGNED-BYTE 32).
2107 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2108 * When built with the :SB-FUTEX feature, threaded builds now take
2109 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2110 for faster/more reliable mutex and condition variable support.
2111 * Incompatible change (but one you probably shouldn't have been using
2112 anyway): the interface and code for arbitrating between multiple
2113 threads in the same user session has been redesigned.
2114 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2115 error when called without an explicit environment argument.
2116 (thanks to Nikodemus Siivola)
2117 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2118 confused about their position. (thanks to Adam Warner and Gerd
2120 * bug fix: Pretty printing backquoted forms with unquotations in the
2121 argument list position of various code constructs such as LAMBDA
2122 now works correctly. (reported by Paul Dietz)
2123 * bug fix: Pretty printing unquotations no longer loses all
2124 stream position information.
2125 * optimization: performance of string output streams is now less
2126 poor for multiple small sequence writes.
2127 * optimization: performance of CSUBTYPEP in the presence of complex
2128 expressions involving CONS and NOT many times has been improved.
2129 (reported by Paul Dietz)
2130 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2132 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2133 streams for the REPL, for the debugger, and so on. (thanks to
2135 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2136 * fixed some bugs revealed by Paul Dietz' test suite:
2137 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2138 is ignored if it is too small, rather than propagating through
2139 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2140 ** extremely complex negations of CONS types were not being
2141 sufficiently canonicalized, leading to inconsistencies in
2143 ** VALUES tranformer lost derived type.
2145 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2146 * fixed a bootstrapping bug: the build process no longer assumes
2147 that the various BOOLE-related constants have the same value in
2148 host and target lisps. (noted by Paul Dietz' test suite on an
2149 SBCL binary built from CLISP)
2150 * The system can now be dynamically linked on the MIPS platform,
2151 which enables dynamic loading of foreign code from Lisp. (thanks
2152 to Ralf Baechle for discussions on the MIPS ABI)
2153 * The system now records debugging information for its own source
2154 files in a filesystem-position-independent manner, relative to
2155 the "SYS" logical host.
2156 * fixed a compiler bug: MV-LET convertion did not check references
2157 to the "max args" entry point. (reported by Brian Downing)
2158 * tweaked disassembly notes to be less confident about proclaiming
2159 some instruction as an LRA. (thanks to Brian Downing)
2160 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2161 now each have their own history, command character, and other
2162 characteristics. (thanks to David Lichteblau)
2163 * fixed some bugs revealed by Paul Dietz' test suite:
2164 ** compiler failure in compiling LOGAND expressions including a
2166 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2167 shift greater than 32.
2168 ** FLUSH-DEST did not mark blocks for type check regeneration.
2169 ** HANDLER-CASE failed to accept declarations in handler clauses
2170 in some circumstances.
2172 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2173 * New code in contrib/sb-introspect (still probably not entirely
2174 stable yet) provides some support for smart Lisp development
2175 environments like SLIME.
2176 * The conditions signalled for errors occurring when loading .fasl
2177 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2178 in a way which should help ASDF recover gracefully.
2179 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2180 stuff has been cleaned up. If you code contrib/ stuff, this might
2181 affect you, and you can look at contrib/README, contrib/STANDARDS,
2182 and/or the 0.8.4.27 diff to check.
2183 * In full calls the compiler now does not generate checks for declared
2184 argument types for all arguments.
2185 * various threading fixes
2186 ** and some experimental patches which didn't make it into
2187 the main tree for this release, but which are shipped in
2188 contrib/experimental-thread.patch as a possible fix for some
2189 failures (deadlock, spinning...) in GC-intensive multithreaded
2191 * fixed PPC build problem (source code incompatibility of different
2192 library versions): added offsetof() hackery which attempts to divine
2193 where glibc maintainers put uc_mcontext today
2194 * fixed bug 282: compiler does not trust type assertions while passing
2195 arguments to a full call.
2196 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2197 &OPTIONAL VALUES type parameter.
2198 * fix bug 214: algorithm for noting rejected templates is now more
2199 similar to that of template seletion. (also reported by rydis on
2201 * fixed bug 141b: printing backquoted information readably and prettily
2202 inserts a space where necessary.
2203 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2204 (CHARACTER 10) are now reported as errors, rather than propagated
2205 as unknown types. (reported by piso on #lisp)
2206 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2207 with values NIL and :ERROR. (thanks to Milan Zamazal)
2208 * fixed bug 191c: CLOS now does proper keyword argument checking as
2209 described in CLHS 7.6.5 and 7.6.5.1.
2210 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2211 longer raise an error; further, using a list as a for-as-arithmetic
2212 counter now raises a meaningful error.
2213 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2214 now understood by sequence creation functions such as MAKE-SEQUENCE
2216 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2217 a STRING-INPUT-STREAM.
2218 * compiler enhancement: SIGNUM is now better able to derive the type
2220 * type declarations inside WITH-SLOTS are checked. (reported by
2222 * fixed some bugs revealed by Paul Dietz' test suite:
2223 ** incorrect optimization of TRUNCATE for a positive first
2224 argument and negative second.
2225 ** compiler failure in let-convertion during flushing dead code.
2226 ** compiler failure while deriving type of TRUNCATE on an
2227 interval, containing 0.
2228 ** ASH of a negative bignum by a negative bignum count now returns
2230 ** intersection of CONS types now canonicalizes properly, fixing
2231 inconsistencies in SUBTYPEP.
2233 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2234 * incompatible change: The --disable-debugger command line
2235 option now clobbers the debugger at a more fundamental
2236 level, by redefining #'INVOKE-DEBUGGER instead of by
2237 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2238 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2239 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2240 enter the debugger REPL and then suffer recursive errors
2241 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2242 while under the new system, BREAK in --disable-debugger mode
2243 terminates the system just as an unhandled error would.
2244 * fixed compiler performance when processing loops with a step >1;
2245 * bug fix: DOCUMENTATION now retrieves generic function
2246 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2247 support has been systematized, and now supports the methods
2248 specified by ANSI, along with a default method and a method for
2249 slot documentation. (reported by Nathan Froyd)
2250 * bug fix: effective methods associated with a generic function are
2251 no longer cached over a change of that generic function's method
2252 combination. (reported by Andreas Fuchs)
2253 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2254 names a non-existent directory. (thanks to Andreas Fuchs)
2255 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2256 the PPC platform, lead to stack corruption; this has been fixed.
2257 (reported by Rainer Joswig)
2258 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2259 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2260 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2261 arithmetic in general, is now much more reliable on the PPC
2263 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2264 platform now returns the right answer.
2265 * optimization: restored some effective method precomputation in
2266 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2267 precomputation is now tunable.
2268 * optimization: compiler-internal data structure use has been
2269 reviewed, and changes have been made that should improve the
2270 performance of the compiler by about 20%.
2271 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2272 simple-base-strings and simple-bit-vectors is improved.
2273 * optimization: the optimization of 32-bit logical and arithmetic
2274 functions introduced in version 0.8.3 on the x86 has been
2275 implemented on the mips, ppc and sparc platforms; an
2276 implementation of the same facility, but for 64-bit arithmetic,
2277 has been added for the alpha.
2278 * microoptimization: the compiler is better able to make use of the
2279 x86 LEA instruction for multiplication by constants.
2280 * bug fix: in some situations compiler did not report usage of
2281 generic arithmetic in (SPEED 3) policy.
2282 * bug 145b fix: compiler used wrong type specifier while converting
2283 MEMBER-types to numeric.
2284 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2286 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2287 index argument now works properly on the Alpha platform.
2288 * bug fix: floating point exception treatment on the Alpha platform
2290 * bug fix: FILE-POSITION works much better on string input and
2291 output streams. (thanks to Nikodemus Siivola)
2292 * bug fix: many threading/garbage collection symptoms sorted.
2293 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2294 might be pseudo-atomic.
2295 * internal change: Stopping for GC is now done with signals not
2296 ptrace. GC is now done in whichever thread wanted it, instead of
2298 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2300 * bug fix: result form in DO is not contained in the implicit
2302 * incompatible change: ICR structure is changed; the value part of
2303 CONTINUATION is now called LVAR; corresponding functions are
2304 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2305 * added type deriver for ISQRT (thanks to Robert E. Brown).
2306 * bug fix: better support for loading from the command line when an
2307 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2309 * fixed some bugs revealed by Paul Dietz' test suite:
2310 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2311 ** ROUND and FROUND now give the right answer when given very
2312 small float arguments.
2313 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2315 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2316 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2317 ** LCM should return a non-negative integer.
2318 ** PARSE-INTEGER returned the index of a terminator instead of the
2319 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2320 ** PARSE-INTEGER returned an incorrect index being applied to a
2322 ** LCM with two arguments of 0 returns 0 rather than signalling
2324 ** unsigned addition of a 32-bit constant with the high bit set no
2325 longer causes an internal compiler error on the x86.
2326 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2327 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2328 with negative last argument.
2329 ** byte specifiers with zero size and position no longer cause
2330 an error during type derivation.
2331 ** bignum multiplication on the Alpha platform now returns the
2333 * porting: The system now builds on SuSE AMD64, although it still
2334 generates a 32-bit binary.
2335 * .fasl file incompatibility: The fasl file version number has
2336 been incremented (because of the changes to internal compiler
2337 data structures referred to above).
2339 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2340 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2341 more accurately, on the Darwin kernel running on PowerPC hardware.
2342 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2343 * Compiler code deletion notes now signal a condition of type
2344 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2345 an associated MUFFLE-WARNING restart.
2346 * The compiler now performs limited argument count validation of
2347 constant format strings in FORMAT, and where appropriate in ERROR,
2348 CERROR and WARN. (thanks to Gerd Moellmann)
2349 * New ASDF-INSTALL contrib can be used for automatic download and
2350 installation of third-party Lisp code from CCLAN or other sites
2352 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2353 which forces another thread to execute a function supplied by the
2355 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2356 now accept and act upon their :ELEMENT-TYPE keyword argument.
2357 (reported by Martin Atzmueller, Edi Weitz)
2358 * bug fix: FILE-POSITION now accepts position designators up to
2359 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2360 is the greater. (thanks to Patrik Nordebo)
2361 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2362 Kalle Olavi Niemitalo)
2363 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2364 printable. (reported by Eric Marsden)
2365 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2366 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2367 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2368 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2369 (thanks to Stig Sandoe)
2370 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2371 UNREAD-CHAR now correctly includes the unread character in the
2372 target sequence. (thanks to Gerd Moellmann)
2373 * bug fix in threaded builds: the system can now be suspended and
2374 resumed by shell job control with minimal disruption.
2375 * bug fixes in times and timezones >2038AD
2376 * better handling of "where is GNU make?" problem in build scripts
2377 (thanks to Nikodemus Siivola)
2378 * new optimization: inside a named function any reference to a
2379 function with the same name is considered to be a self-reference;
2380 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2381 optimization quality.
2382 * new optimization on x86: logical functions and + now have
2383 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2384 used when the result is truncated to 32 bits.
2385 * VALUES declaration is partially enabled.
2386 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2387 * fixed some bugs revealed by Paul Dietz' test suite:
2388 ** The system now obeys the constraint imposed by
2389 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2390 types form a lattice under type intersection.
2391 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2392 ** ASSOC now ignores NIL elements in an alist.
2393 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2394 and (1+ MOST-POSITIVE-FIXNUM) answers.
2395 ** The addition of a method with invalid qualifiers to a generic
2396 function does not cause an error to be signalled immediately;
2397 a warning is signalled, and the error is generated only on
2398 calling the generic function.
2399 * changed .fasl file version number, in part to add type codes for
2400 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2401 obscure ANSI requirements
2403 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2404 * fixed bug 148: failure to inline-expand a local function left
2405 garbage, confusing the compiler.
2406 * fixed bugs 3cd: structure slot readers perform type check if the
2407 slot can have an invalid value (i.e. it is either not initialized
2408 or can be written with a less specific slot writer).
2409 * bug fix: the compiler now traps array references to elements off
2410 the end of an array; previously, the bounds checking in some
2411 circumstances could go off-by-one.
2412 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2414 * type declarations for array element types now obey the description
2415 on the CLHS page "Declaration TYPE", as per discussions on
2416 sbcl-help around 2003-05-08. This means that a declaration
2417 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2418 declaration, all references to BAR will be asserted or assumed
2419 (with THE, so dependent on compiler policy) to involve objects of
2420 type FOO. Note that no such declaration is implied in
2421 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2422 * declared types of functions from the "Conditions"
2423 chapter. (reported by Paul Dietz)
2424 * bug fix: CERROR accepts a function as its first argument.
2425 * bug fix: NTH an NTHCDR accept a bignum as index
2426 arguments. (reported by Adam Warner)
2427 * optimization: character compare routines now optimize comparing
2428 against a constant character. (reported by Gilbert Baumann)
2429 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2430 argument now works properly on the MIPS platform.
2431 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2432 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2433 declared to have a functional type, does not cause an error
2435 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2436 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2437 * fixed bug: if last continuation of a deleted block has a
2438 destination, this destination should be deleted too. (reported by
2440 * fixed a bug in the bootstrap process: the host compiler's values
2441 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2442 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2443 test case from Patrik Nordebo)
2444 * improved the ability of the disassembler on the PPC platform to
2445 provide helpful disassembly notes.
2446 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2447 the class in more cases than previously.
2448 * bug fix: FILE-POSITION now understands :START and :END for
2449 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2450 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2451 8) data. (thanks to David Lichteblau)
2452 * bug fix: it is possible to add a method to a generic function
2453 without lambda list.
2454 * bug fix: reader failed to signal END-OF-FILE inside an
2455 object representation. (reported by Nikodemus Siivola)
2456 * fixed some bugs revealed by Paul Dietz' test suite:
2457 ** LAST and [N]BUTLAST should accept a bignum.
2458 ** condition slot accessors are methods.
2459 ** (VECTOR NIL) is a subtype of STRING.
2461 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2462 * minor incompatible change: some nonsensical specialized lambda
2463 lists (used in DEFMETHOD) which were previously ignored now signal
2465 * minor incompatible change: the system is now aware of the types of
2466 variables in the COMMON-LISP package, and will signal errors for
2467 most violations of these type constraints (where previously they
2468 were silently accepted).
2469 * minor incompatible change: COMPILE-FILE now uses the freedom
2470 afforded (ANSI 3.2.2.3) to use derived function types for
2471 functions defined in the same file. This also permits the system
2472 to warn on static type mismatches and function
2473 redefinition. (Currently it does not work with high DEBUG level.)
2474 * minor incompatible change: VALUES declaration is disabled.
2475 * When issuing notes, the compiler now signals a condition of type
2476 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2477 restart for use in user handlers. It is expected that the
2478 COMPILER-NOTE condition will eventually become a condition
2479 supertype to a hierarchy of note types, which will then be
2480 handleable in a similar fashion. However, at the moment, no such
2481 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2482 but it's an implementation detail, not a classification for the
2484 * Changes in type checking closed the following bugs:
2485 ** type checking of unused values (192b, 194d, 203);
2486 ** template selection based on unsafe type assertions (192c, 236);
2487 ** type checking in branches (194bc).
2488 * A short form of VALUES type specifier has ANSI meaning (it has
2489 increased the number of situations when SBCL cannot perform type
2491 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2492 keywords or constants is permissible.
2493 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2494 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2495 classes. (thanks to Antonio Martinez)
2496 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2497 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2498 argument to a pretty-print stream. Instead, it leaves any such
2499 operations to DESCRIBE-OBJECT methods.
2500 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2502 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2503 lambda lists. (thanks to David Lichteblau)
2504 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2505 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2506 (thanks to Gerd Moellmann)
2507 * fixed bug 246: increased compilation speed of long
2508 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2510 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2512 * DEFCONSTANT now throws a condition of type
2513 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2514 constant to a non-EQL value; CONTINUE and ABORT restarts
2515 respectively change and preserve the value.
2516 * fixed bug 63: The code walker, part of the implementation of CLOS,
2517 is now better at handling symbol macros.
2518 * bug fix: There is no longer an internal implementation type named
2519 CL:LENGTH. (reported by Raymond Toy)
2520 * bug fix: In macro-like defining macros/special operators the
2521 implicit block does not enclose the lambda list.
2522 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2523 enclosed in parentheses) are not suitable as type specifiers, and
2524 their use properly signals an error now.
2525 * bug fix: An argument count mismatch for a type specifier in code
2526 being compiled no longer causes an unhandled error at compile
2527 time, but signals a compile-time warning.
2528 * fixed simple vector readable printing
2529 * bug fix: DESCRIBE takes more care over whether the class
2530 precedence list slot of a class is bound before accessing it.
2531 (reported by Markus Krummenacker)
2532 * bug fix: FORMATTER can successfully compile pretty-printer format
2533 strings which use variants of the ~* directive inside.
2534 * bug fix: SEARCH now applies its TEST predicate to the elements of
2535 the arguments in the correct order. (thanks to Wolfhard Buss)
2536 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2537 in type inference. (thanks to Robert E. Brown)
2538 * bug fix: Reading in symbols with an explicit package name of ""
2539 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2540 * fixed some bugs revealed by Paul Dietz' test suite:
2541 ** NIL is now allowed as a structure slot name.
2542 ** Arbitrary numbers, not just REALs, are allowed in certain
2543 circumstances in LOOP for-as-arithmetic clauses.
2544 ** Multiple class redefinitions before slot access no longer
2545 causes a type error.
2546 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2547 association between the name and a class.
2548 ** Generic functions with non-standard method-combination and over
2549 five methods all of which return constants no longer return NIL
2550 after the first few invocations. (thanks to Gerd Moellmann)
2551 ** CALL-NEXT-METHOD with no arguments now passes the original
2552 values of the arguments, even in the presence of assignment.
2553 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2555 ** DISASSEMBLE works with closures and funcallable instances.
2556 ** ADD-METHOD now returns the generic function, not the new method.
2557 ** FIND-METHOD signals an error if the lengths of the specializers
2558 is incompatible with the generic function, even if the ERRORP
2560 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2561 which its argument is a member.
2562 ** DEFCLASS only redefines the class named by its class-name
2563 argument if that name is the proper name of the class;
2564 otherwise, it creates a new class.
2565 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2566 of the UNBOUND-SLOT condition to the name of the slot.
2567 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2568 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2569 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2570 SLOT-MAKUNBOUND in the specified fashion.
2572 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2573 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2574 cross-compilation host. As a consequence, we can now bootstrap our
2575 way up to SBCL starting with a bare gcc toolchain and human-readable
2576 source code (first the source to CLISP, then the source to SBCL).
2577 * A contributed module containing a partial implementation of the
2578 simple-streams interface has been included. (thanks to Rudi
2580 * A contributed module implementing the RFC1321 Message Digest
2581 Algorithm, known as MD5, has been included.
2582 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2583 no longer has any effect, as the code controlled by this feature
2584 has been deleted. (As far as we know, no-one has ever built using
2585 this feature, and its semantics were confused in any case).
2586 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2587 work on conditions (as required by the ANSI specification),
2588 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2589 expected behaviour on conditions. Users should note, however,
2590 that such behaviour is not required by the ANSI specification,
2591 and so use of this behaviour may render their code unportable.
2592 * fixed some bugs revealed by Paul Dietz' test suite:
2593 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2595 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2596 ** SXHASH on condition objects no longer returns NIL.
2597 ** :ALLOCATION :CLASS slots are better treated; their values are
2598 updated on class redefinition, and initforms inherited from
2599 superclasses are applied.
2600 ** REMOVE-METHOD returns its generic function argument even when
2601 no method was removed.
2602 ** SHARED-INITIALIZE now initializes the values of the requested
2603 slots, including those with :ALLOCATION :CLASS.
2604 ** ALLOCATE-INSTANCE now works on structure classes defined via
2605 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2607 ** SLOT-EXISTS-P now works on conditions, as well as structures
2609 ** MAKE-LOAD-FORM now has the required methods on
2610 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2611 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2612 keyword, and now implements the SLOT-NAMES argument.
2613 ** methods with &OPTIONAL arguments no longer allow too many
2614 arguments to be passed in the call without error.
2615 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2616 option is consistent with the required arguments of the generic
2617 function lambda list.
2618 * bug fix: REQUIRE accepts a string designator. (Thanks to
2620 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2621 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2622 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2623 * bug fix: The compiler now checks for duplicated variables in macro
2625 * bug fix: SETQ on globals returns the correct value.
2626 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2627 not a COMPILER-ERROR (followed by some other strange error on
2628 choosing the CONTINUE restart).
2629 * bug fix: make.sh and friends are now more consistent in the way that
2630 they look for GNU "make".
2632 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2633 * experimental native threads support (on x86 Linux >=2.4 only).
2634 This is not compiled in by default: you need to add :SB-THREAD to
2635 the target features. See the "Beyond ANSI" chapter of the manual
2637 * fix for longstanding nonANSIism: The old distinction between
2638 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2639 The return value from CL:FIND-CLASS is now a CLOS class, and
2640 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2641 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2642 * An interface to the MetaObject Protocol, as described in Kiczales,
2643 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2644 MIT Press, 1991, is available from the SB-MOP package.
2645 * incompatible change: the SB-PCL package should now be considered
2646 a private implementation detail, and no longer a semi-private MOP
2648 * minor incompatible change: due to rearrangement for threads, the
2649 control stack and binding stack are now allocated at arbitrary
2650 addresses instead of being hardcoded per-port. Users affected by
2651 this probably have to be doing advanced things with shared
2652 libraries, and will know who they are.
2653 * minor incompatible change: Previously, all --eval forms used were
2654 processed with READ before any of them were processed with EVAL.
2655 Now each --eval form is processed with both READ and EVAL before
2656 the next --eval form is processed. (Thus package operations like
2657 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2658 work as the user might reasonably expect.)
2659 * minor incompatible change: *STANDARD-INPUT* is now only an
2660 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2662 * minor incompatible change: Y-OR-N-P is now character-oriented,
2663 not line oriented. Also, YES-OR-NO-P now works without errors.
2664 (thanks to Antonio Martinez)
2665 * sb-aclrepl module improvements: an integrated inspector, added
2666 repl features, and a bug fix to :trace command.
2667 * Known functions, which cannot be open coded by the backend, are
2668 considered to be able to check types of their arguments. (fixing
2669 a bug report by Nathan J. Froyd)
2670 * fixed a bug in computing method discriminating functions: It is
2671 now possible to define methods specialized on classes which have
2672 forward-referenced superclasses. (thanks to Gerd Moellmann)
2673 * fixed evaluation order in optional entries (reported by Gilbert
2675 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2676 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2677 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2678 required-by-AMOP NAME argument, as well as CLASS and
2679 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2680 * fixed bug 20: DEFMETHOD can define methods using names that are
2681 not the proper names of classes to designate class specializers.
2682 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2683 Unix file descriptors, instead of blowing up. (thanks to Antonio
2685 * Garbage collection refactoring: user-visible change is that a
2686 call to the GC function during WITHOUT-GCING will not do garbage
2687 collection until the end of the WITHOUT-GCING. If you were doing
2688 this you were probably losing anyway.
2689 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2690 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2691 TYPEP the latter but not the former.
2692 * The compiler issues a full WARNING for calls to undefined functions
2693 with names from the CL package.
2694 * MAP-INTO for a vector destination is open coded. (reported by
2695 Brian Downing on c.l.l)
2696 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2697 documentation string.
2698 * fixed some bugs revealed by Paul Dietz' test suite:
2699 ** COPY-ALIST now signals an error if its argument is a dotted
2701 ** Condition slots are now accessed more correctly in the presence
2702 of multiple initargs for a given slot.
2703 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2704 exclude restarts of the same name associated with a different
2706 ** DEFCLASS of forward-referenced classes with another
2707 forward-referenced class in the superclasses list no longer
2709 ** Condition slots are now initialized once each, not multiple
2710 times. (thanks to Gerd Moellmann)
2711 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2713 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2714 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2715 arguments contain duplicated elements.
2716 ** RESTART-CASE understands local macros.
2717 ** RESTART-CASE associates exactly its own restarts with a condition.
2718 ** ENDP in safe mode checks its argument to be of type LIST.
2719 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2720 in question is unbound.
2721 ** Optimized MAKE-INSTANCE functions no longer cause internal
2722 assertion failures in the presence of duplicate initargs.
2723 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2724 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2725 propagated even in the ctor optimized implementation of
2727 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2729 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2730 * a better implementation of SXHASH on (simple) bit vectors,
2731 measured both in execution speed and in distribution of results
2732 over the positive fixnums, has been installed. Likewise, a better
2733 implementation of EQUAL for simple bit vectors is now available.
2734 * fixed CEILING optimization for a divisor of form 2^k.
2735 * fixed bug 240 (emitting extra style warnings "using the lexical
2736 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2737 by Antonio Martinez)
2738 * fixed SXHASH, giving different results for NIL depending on type
2739 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2740 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2741 variable. (found by Rolf Wester)
2742 * SBCL does not ignore type declarations for special
2743 variables. (reported by rif on c.l.l 2003-03-05)
2744 * some bug fixes in contrib/sb-aclrepl/
2745 * fixed some bugs revealed by Paul Dietz' test suite:
2746 ** a bug in the CONS type specifier, whereby the CAR and CDR
2747 types got intertwined, has been fixed;
2748 ** the type system is now able to reason about the interaction
2749 between INTEGER and RATIO types more completely;
2750 ** APPEND, [N]REVERSE and NRECONC check that those their
2751 arguments, which must be proper lists, are really so;
2752 ** An array specialized to be unable to hold elements has been
2753 implemented, as required -- yes, really -- by ANSI;
2754 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2755 on malformed property lists;
2757 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2758 * incompatible packaging change: in line with Unix convention,
2759 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2760 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2761 * REQUIRE and PROVIDE are now optionally capable of doing something
2762 useful. See the documentation string for REQUIRE.
2763 * infrastructure for a managed SBCL contrib system: contributed
2764 modules in this release include:
2765 ** the ASDF system definition facility;
2766 ** an interface to the BSD Sockets API;
2767 ** an ACL-like convenience interface to the repl;
2768 (thanks to Kevin Rosenberg)
2769 ** an implementation of ROTATE-BYTE, with efficient implementation
2771 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2772 gives a closer approximation to the right answer than previously.
2773 (thanks to Raymond Toy)
2774 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2775 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2776 optional environment arguments, as required by ANSI.
2777 * fixed bugs in other functions taking environment objects, allowing
2778 calls with an explicit NIL environment argument to be compiled
2780 * fixed bug 228: primary return values from
2781 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2782 COMPILE or FUNCTION.
2783 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2784 :TYPE will no longer signal errors on innocuous objects.
2785 * fixed bug 231b: SETQ is better at respecting type declarations in
2786 the lexical environment.
2787 * fixed a bug in DEFCLASS: classes named by symbols with no or
2788 unprintable packages can now be defined.
2789 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2790 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2791 * fixed compiler failure related to checking types of functions.
2792 (reported by Robert E. Brown)
2793 * the compiler is now much more consistent in its error-checking
2794 treatment of bounding index arguments to sequence functions: in
2795 (SAFETY 3) code, errors will be signalled in almost all cases if
2796 invalid sequence bounding indices are passed to functions defined
2797 by ANSI to operate on sequences.
2798 * fixed a bug in the build procedure: documentation of SBCL-specific
2799 packages is now preserved and available in the final Lisp image.
2800 * lifted FDEFINITION lookup out of loops in the implementation of
2801 many list operations. (thanks to Robert E. Brown)
2802 * fixed a bug in the reader: the #n# reader macro now works for
2803 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2804 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2805 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2806 E. Brown and Tony Martinez respectively)
2807 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2808 SLOT-BOUNDP: in particular, it is now able to optimize them much
2809 better, and is now not vulnerable to having packages renamed.
2810 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2811 always called when a slot is not present in an instance. (thanks
2813 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2814 longer signal an error on unreading a character following EOF on
2815 the previous constituent stream. (thanks to Tony Martinez)
2816 * fixed some bugs revealed by Paul Dietz' test suite:
2817 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2818 not just nonnegative fixnums;
2819 ** the logical bit-array operators such as BIT-AND now accept an
2820 explicit NIL for their "opt-arg" argument (to indicate a
2821 freshly-consed result bit-array);
2822 ** ELT now signals an error on an invalid sequence index in safe
2824 ** the type system is now cleverer about negations of numeric
2825 types, and consequently understands the BIGNUM and RATIO types
2827 ** the type system is now cleverer about the interaction between
2828 INTEGER and RATIO types: while bugs still remain, many more
2829 cases are accurately computed;
2830 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2831 if it is in the last clause;
2832 ** CONSTANTLY now correctly returns a side-effect-free function in
2834 ** DECLARE is no longer treated as a special-operator; in
2835 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2836 * incremented fasl file version number due to the change in the
2837 DEFSTRUCT-SLOT-DESCRIPTION structure.
2839 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2840 * minor incompatible change: code processed by the "interpreter" or
2841 EVAL now has a compilation optimization policy of (DEBUG 2)
2842 (changed from (DEBUG 1)) to improve debuggability of interactive
2843 development, and to allow the use of the debug RETURN command in
2845 * an experimental implementation of the RETURN command for the
2846 debugger has been included. (thanks to Frederik Kuivinen)
2847 * fixed bug 62: constraints were not propagated into a loop.
2848 * fixed bug in embedded calls of SORT (reported and investigated by
2850 * fixed some bugs revealed by Paul Dietz' test suite:
2851 ** printing and reading of arrays with some dimensions having
2852 length 0 (thanks to Gerd Moellmann);
2853 ** BOA constructor with &AUX argument without a default value does
2854 not cause a type error;
2855 ** CONSTANTP now returns true for all self-evaluating objects.
2857 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2858 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2859 accessors that are related by inheritance, as specified in the
2860 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2861 Valtteri Vuorikoski)
2862 * The compiler is now able to inline functions that were defined in
2863 a complex lexical environment (e.g. inside a MACROLET).
2864 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2866 * Support for the upcoming FreeBSD-5.0 release has been included.
2867 (thanks to Dag-Erling Smorgrav)
2868 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2869 effect when it is not in a toplevel context.
2870 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2871 stand a better chance of being correct. (thanks to Gerd
2873 * fixed bug in COERCE, which now signals an error on coercing a
2874 rational to a bounded real type which excludes the expected
2876 * The compiler is now able to derive types more accurately from the
2877 COERCE and COMPILE functions.
2878 * fixed bug 223: functional binding is considered to be constant
2879 only for symbols in the CL package.
2880 * fixed bug 231: SETQ did not check the type of a variable being set
2881 (reported by Robert E. Brown)
2882 * A new optimization for MAKE-INSTANCE has been included, fixing
2883 various bugs (including relating to :ALLOCATION :CLASS slots and
2884 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2886 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2887 ** As required by ANSI, LOOP now disallows anonymous collection
2888 clauses such as COLLECT I in conjunction with aggregate boolean
2889 clauses such as THEREIS (= I 1);
2890 ** LOOP now signals an error when any variable is reused in the
2891 same loop (including the potentially useful construct analogous
2892 to WITH A = 1 WITH A = (1+ A);
2893 ** IT is only a special loop symbol within the first clause of a
2894 conditional loop clause;
2895 ** LOOP with a typed iteration variable over a hashtable now
2896 signals a type error iff it should.
2897 * fixed some other bugs revealed by Paul Dietz' test suite:
2898 ** FILE-STREAM now names the class previously known as FD-STREAM;
2899 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2900 argument) no longer signals an error;
2901 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2902 of the slot symbol, rather than using the current package
2903 ((:CONC-NAME "") continues to intern the slot's name in the
2905 * incremented fasl file version number, because of the incompatible
2906 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2907 of the new implementation of DEFINE-COMPILER-MACRO.
2909 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2910 * Support for building SBCL for MIPS platforms running in
2911 little-endian mode has now been checked in, and basic
2912 functionality on said platforms verified.
2913 * minor incompatible change: PCL now records the pathname of a file
2914 in which methods and the like are defined, rather than its
2916 * minor incompatible change: TRUENAME now considers the truename of
2917 a file naming a directory to be the pathname with :DIRECTORY
2918 component indicating that directory.
2919 * minor incompatible change: a NAMED clause in the extended form of
2920 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2921 reason for the previous behaviour is unclear.
2922 * more systematization and improvement of CLOS and MOP conformance
2923 in PCL (thanks to Gerd Moellman and Pierre Mai):
2924 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2926 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2927 primary methods with no specializers;
2928 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2930 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2931 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2932 FORWARD-REFERENCED-CLASSes; error reporting on
2933 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2935 ** SXHASH on CLOS instances now uses a slot internal to the
2936 instance to return different numbers on distinct instances,
2937 while preserving the same return value through invocations of
2939 ** DEFMETHOD signals errors when methods with longer incongruent
2940 lambda lists are added to generic functions;
2941 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2942 CLASS, as specified in AMOP;
2943 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2944 on the order returned by the primary method for classes of
2945 class STANDARD-CLASS;
2946 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2947 * fixed some bugs shown by Paul Dietz' test suite:
2948 ** DOLIST puts its body in TAGBODY;
2949 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2951 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2952 value producing form;
2953 ** if more variables are given to PROGV than values, extra
2954 variables are bound and made to have no value;
2955 ** NSUBSTITUTE on list arguments gets the right answer with
2957 ** ELT signals an error of type TYPE-ERROR when the index argument
2958 is not a valid sequence index;
2959 ** LOOP signals (at macroexpansion time) an error of type
2960 PROGRAM-ERROR when duplicate variable names are found;
2961 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2962 ** FUNCALL of special-operators now cause an error of type
2964 ** PSETQ now works as required in the presence of side-effecting
2965 symbol-macro places;
2966 ** NCONC accepts any object as its last argument;
2967 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2969 ** loop-for-as-package does not require a package to be explicitely
2971 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2973 * fixed bug 166: compiler preserves "there is a way to go"
2974 invariant when deleting code.
2975 * fixed bug 172: macro lambda lists with required arguments after
2976 &REST arguments now cause an error to be signalled. (thanks to
2978 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2979 bug. (thanks to Matthew Danish)
2980 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2982 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2983 arguments are lexically rebound. (thanks to Gerd Moellmann and
2985 * fixed bug 194: error messages are now more informative when there
2986 is no primary method applicable in a call to a generic
2987 function. (thanks to Gerd Moellmann)
2988 * fixed bug in command line argument checking (thanks to Julian
2990 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2991 (thanks to Matthew Danish)
2992 * incremented fasl file version number, because of the
2993 SXHASH-related changes in the layout of CLOS data structures
2995 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2996 * minor incompatible change: The runtime (the Unix executable named
2997 "sbcl") is now much pickier about the .core files it will load.
2998 Essentially it now requires .core files to descend from the same
2999 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3000 as the runtime does. (The intent is to prevent the crashes which
3001 can occur, and which can even be reported as mysterious failures,
3002 when people patch the sources or change the build parameters
3003 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3004 sbcl and .core files.)
3005 * fixed bug: VALUES-LIST is no longer optimized away.
3006 * fixed bug 142: The FFI conversion of C string values to Lisp
3007 string values no longer conses excessively. (thanks to Nathan
3008 Froyd porting Raymond Toy's fix to CMU CL)
3009 * began to systematize and improve MOP conformance in PCL (thanks to
3010 Nathan Froyd, Gerd Moellman and Pierre Mai):
3011 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3013 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3014 ** FINALIZE-INHERITANCE is now called on class finalization;
3015 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3016 argument precedence order.
3017 * fixed bug 202: The compiler no longer fails on functions whose
3018 derived types contradict their declared type.
3019 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3020 so it can be non-toplevel.
3021 * The fasl file version number has changed (because of the new
3022 implementation of DEFMACRO).
3023 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3024 safe code, that any length requirement by their type-specifier
3025 argument is valid. The exceptions to this are described in bug
3027 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3028 functions now check the types of their inputs as required by ANSI.
3029 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3030 introduced symbol is DECLAREd to be SPECIAL.
3031 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3032 * fixed bug: :COUNT argument to sequence functions may be negative
3033 * fixed bug: body of DO-SYMBOLS may contain declarations
3034 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3035 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3036 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3037 symbol macro only once
3038 * fixed printing of call frame when argument list is unavailable
3039 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3040 * compiler no longer signals WARNING on unknown keyword
3043 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3044 * A beta-quality port to the mips architecture running Linux,
3045 based on the old CMUCL backend, has been made. It has been tested
3046 on a big-endian kernel, and works sufficiently well to be able to
3047 rebuild itself; it has not been tested in little-endian mode.
3048 * fixed an inconsistency between gencgc.c and purify.c which made
3049 dumping/loading .core files unreliable
3050 * fixed bug 120a: The compiler now deals correctly with IFs where
3051 the consequent is the same as the alternative, instead of
3052 misderiving the return type. (thanks to Alexey Dejneka)
3053 * fixed bug 113: Logical pathnames are now dumpable (the logical
3054 host is resolved at load-time, throwing an error if it is not
3056 * fixed bug 174: FORMAT's error message is slightly clearer when a
3057 non-printing character is used in a format directive.
3058 * fixed several bugs in compiler checking of type declarations, i.e.
3059 violations of the Python "declarations are assertions" principle
3060 (thanks to Alexey Dejneka)
3061 * fixed several bugs in PCL's error checking (thanks to Gerd
3063 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3065 * fixed bug in compilation of functions as first class values
3066 (thanks to Antonio Martinez-Shotton)
3067 * The compiler's handling TYPE-ERRORs which it can prove will
3068 inevitably happen at runtime has been cleaned up and corrected
3069 in several ways. (thanks to Alexey Dejneka)
3070 * improved argument type checking for various basic arithmetic
3071 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3072 lavished on them in the past that they can be compiled in many
3073 ways in different special cases
3074 * fixed bug 181: compiler checks validity of user supplied type
3076 * cleaned up code flushing in optimization: Function calls which
3077 should signal errors for safety purposes (e.g. which ANSI says
3078 should signal errors when their arguments are of incorrect type)
3079 are no longer optimized away.
3080 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3081 * incremented fasl file version number, because changes in the
3082 implementation of sequence functions like COERCE caused
3083 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3084 in old inline expansions) to become undefined. (Actually these
3085 changes were later undone, so we might very well be binary
3086 compatible with 0.7.7 after all, but leaving the version number
3087 incremented seemed like the simplest and most conservative
3090 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3091 * An alpha-quality port to the parisc architecture running Linux,
3092 based on the old CMUCL backend, has been made. This, even more so
3093 than the other backends, should be considered still a work in
3094 progress; known problems include that the Linux kernel in 64-bit
3095 mode does not propagate the correct sigcontext structure to
3096 userspace, and consequently SBCL on a parisc64 kernel will not
3098 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3099 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3100 suppressed.) Also now that the compiler is looking at declarations
3101 in the environment, it checks optimization declarations as well,
3102 and suppresses inlining when (> DEBUG SPEED).
3103 * More fixes have been made to treatment of floating point exception
3104 treatment and other Unix signals. In particular, floating point
3105 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3106 * The detection and handling of control stack exhaustion (infinite
3107 or very deeply nested recursion) has changed. Stack exhaustion
3108 detection is now done by write-protecting pages at the OS level
3109 and applies at all optimization settings; when found, a
3110 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3111 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3112 be caught using IGNORE-ERRORS.
3113 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3114 correctly when transforming recursive LABELS functions to LETs.
3115 (thanks to Alexey Dejneka)
3116 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3117 that are names of constants or global variables.
3118 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3119 alien routines with docstrings.
3120 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3121 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3123 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3124 (thanks to Eric Marsden)
3125 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3126 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3127 * Bug fix: The SPARC backend can now compile functions involving
3128 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3129 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3130 to a functions expecting a PATHNAME-DESIGNATOR.
3131 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3132 lambda lists. (thanks to Alexey Dejneka)
3133 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3134 functions so that EXPORTing the name of the class doesn't cause
3135 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3136 to Antonio Martinez for reporting this)
3137 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3138 constant-folded without failing an assertion. (thanks to Einar
3139 Floystad Dorum for reporting this)
3140 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3141 types are now dealt with more correctly by the compiler.
3142 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3143 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3144 behaviour analogous to RENAME-FILE. This puts its behaviour more
3145 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3147 * The fasl file version number has changed again. (because of the
3148 bug fix involving the names of PCL MAKE-INSTANCE functions)
3150 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3151 * bug fix: Floating point exceptions are treated much more
3152 consistently on the x86/Linux and PPC/Linux platforms.
3153 * Array initialization with :INITIAL-ELEMENT is now much faster for
3154 cases when the compiler cannot open code the array creation, but
3155 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3156 array accesses have also seen a speed increase.
3157 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3158 specified. (This was at the root of some bad interactions between
3159 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3161 * bug fix: Internal error arguments for undefined functions are now
3162 computed correctly on the PPC/Linux platform.
3163 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3164 Raymond Toy's patch for CMU CL)
3165 * Support for the Solaris 9 operating environment has been included
3166 (thanks to Daniel Merritt)
3167 * A very ugly but hopefully complete draft of the missing FFI chapter
3168 of the manual has been created by reformatting the corresponding
3169 CMU CL manual chapter into (currently very ugly and incoherent)
3170 DocBook and bringing it up to date for SBCL behavior. Thus, the
3171 manual is now essentially complete, at least by my extreme
3172 once-and-only-once standards, whereby it's acceptable to refer to
3173 the doc strings of SB-EXT functions as the primary documentation.
3174 * The fasl file version number has changed again, due to cleanup of
3175 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3176 is no longer a static symbol.)
3178 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3179 * SBCL now builds with OpenMCL (version 0.12) as the
3180 cross-compilation host; also, more progress has been made toward
3181 bootstrapping under CLISP.
3182 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3184 * bug 158 fixed: The compiler can now deal with integer loop
3185 increments different from 1; fixing this turned out also to fix
3187 * bug 169 fixed: no more bogus warnings about using lexical bindings
3188 despite the presence of perfectly good SPECIAL declarations (thanks
3189 to David Lichteblau)
3190 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3191 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3192 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3194 * bug 180 fixed: Method combination specifications no longer ignore
3195 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3196 * bug fix: Structure type predicate functions now check their argument
3197 count as they should.
3198 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3199 correctly. (thanks to Pierre Mai)
3200 * minor incompatible change: The --noprogrammer option is deprecated
3201 in favor of the new --disable-debugger option, which is very similar.
3202 (The major difference is that it takes effect at a slightly different
3203 time at startup, causing handling of errors in --sysinit and
3204 --userinit files will be affected differently.) The
3205 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3206 been added to allow this functionality to be controlled from ordinary
3207 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3208 maintainers, who might want to run non-interactive scripts to
3209 build SBCL cores which will later be used interactively.)
3210 * minor incompatible change: The LOAD function no longer, when given
3211 a wild pathname to load, loads all files matching that pathname.
3212 Instead, an error of type FILE-ERROR is signalled.
3214 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3215 * bug 147 fixed: The compiler preserves its block link/count
3216 invariants more correctly now so that it doesn't crash. (thanks
3218 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3220 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3221 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3222 no-value-for-this-character as NIL (as specified by ANSI).
3223 * HOST-NAMESTRING on physical pathnames now returns a string that is
3224 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3225 (thanks to Christophe Rhodes)
3226 * The Alpha port handles icache flushing more correctly. (thanks to
3228 * More progress has been made toward bootstrapping under CLISP. (thanks
3229 to Christophe Rhodes)
3230 * The fasl file format has changed again, because dynamic loading
3231 on OpenBSD (which has non-ELF object files) motivated some cleanups
3232 in the way that foreign symbols are transformed and passed around.
3233 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3234 is no longer treated as whitespace by the reader, but instead as
3235 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3236 |A
\7fB|, instead of A as it used to.
3238 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3239 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3240 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3241 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3242 this as of 0.7.1.45, but was left out of the previous news section
3243 (thanks to Dan Barlow)
3244 * SBCL now runs on the Solaris operating system on SPARC architectures
3245 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3246 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3247 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3248 * SPARC backend cleanups, allowing builds of cores optimized for V8
3249 and V9 SPARCS, and also emission of code targeted to a particular
3250 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3252 * SBCL is closer to bootstrapping under CLISP, thanks to various
3253 fixes by Christophe Rhodes.
3254 * The fasl file format has changed again, to allow the compiler's
3255 INFO database to support symbol macros.
3256 * The user manual (in doc/) is formatted into HTML more nicely.
3257 (thanks to coreythomas)
3258 * The system is smarter about SUBTYPEP relationships, especially
3259 those involving NOT types (including types such as ATOM which are
3260 represented internally using NOT types). Thus SUBTYPEP is less
3261 likely to return (VALUES NIL NIL) in general, and in particular
3262 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3264 * The fasl file format has changed again, because the internal
3265 representation of types now includes a new slot to support the new
3266 SUBTYPEP-of-NOT-types logic.
3267 * (not a change in the main branch of SBCL, but a related prototype
3268 which can hopefully be merged into the main branch of SBCL in the
3269 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3270 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3271 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3272 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3273 dispatch macro character combinations. (thanks to Alexey Dejneka)
3274 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3275 that superficially look like logical namestrings correctly.
3276 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3278 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3279 default supplied-p) for &optional and &key arguments. (thanks to
3282 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3283 * incompatible change: The compiler is now less aggressive about
3284 tail call optimization, doing it only when (> SPACE DEBUG) or
3285 (> SPEED DEBUG). (This is an incompatible change because there are
3286 programs which relied on the old CMU-CL-style behavior to optimize
3287 away their unbounded recursion which will now die of stack overflow.)
3288 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3289 for non-GENCGC systems has been increased to 20M (since that
3290 seems much closer to the likely performance optimum for modern
3291 systems than the old 4M value was)
3292 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3293 and *DEBUG-PRINT-LEVEL*
3294 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3295 of CMU CL's support for SPARC, and various endianness and other
3296 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3297 * new syntactic sugar for the Unix command line: --load foo.bar is now
3298 an alternate notation for --eval '(load "foo.bar")'.
3300 ** The system now detects stack overflow and handles it gracefully,
3301 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3302 optimization settings. (This is a good thing in general, and
3303 its introduction in this version should be particularly timely
3304 for anyone whose code fails because of suppression of tail
3306 ** The system now hunts for the C variable "environ" in a more
3307 devious way, to avoid segfaults when the C library version differs
3308 between compile time and run time. (thanks to Christophe Rhodes)
3309 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3310 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3311 ** The compiler no longer issues bogus style warnings for undefined
3312 classes in the same source file as the DEFCLASSes which defined
3313 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3315 ** fixes in CONDITION class precedence list for undefined function
3316 errors (thanks to Alexei Dejneka)
3317 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3318 correctly. (thanks to Dan Barlow)
3319 ** portability fixes aiming at bootstrapping under CLISP (thanks
3320 to Dave McDonald and Christophe Rhodes)
3321 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3322 ** fixes in type translation and and type inference (thanks to
3324 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3325 ** various fixes in the new ports (thanks to Dan Barlow)
3326 * several changes related to debugging:
3327 ** suppression of tail recursion, as noted above
3328 ** stack overflow detection, as noted above
3329 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3330 is now the default. (For some time encapsulation has been more
3331 reliable than the breakpoint-based :ENCAPSULATE NIL
3332 implementation, at least on X86 systems; and I just noticed that
3333 encapsulation also seems closer to the spirit of the ANSI
3336 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3338 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3339 up properly again. (There was a packaging bug in 0.7.0 which
3340 left their definitions in SB-SYS::LOAD-FOREIGN and
3341 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3342 vital for most things which interface to C-level interfaces,
3343 like extensions working with sockets or databases or
3344 Perl-compatible regexes or whatever, and the need to fix
3345 this bug is the main reason that 0.7.1 was released so
3347 ** DEFGENERIC is now choosier about the methods it redefines, so that
3348 reLOADing a previously-LOADed file containing DEFGENERICs does
3349 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3350 cycle now works as it should. (thanks to Alexey Dejneka)
3351 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3352 fixed by Christophe Rhodes. (He actually submitted this patch
3353 months ago, and I delayed until after 0.7.0.)
3354 ** Bug 111 (internal compiler confusion about runtime checks on
3355 FUNCTION types) was fixed by Alexey Dejneka.
3356 * Some internal cleanups (getting rid of variables which aren't
3357 needed now that the byte interpreter is gone) caused the fasl
3358 file format number to change again.
3360 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3361 * major incompatible change: The default fasl file extension, i.e. the
3362 default extension for files produced by COMPILE-FILE, has changed
3363 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3365 ** There are many changes in the implementation of the compiler.
3366 SBCL is now essentially a compiler-only implementation of ANSI
3367 Common Lisp. EVAL still "interprets" a few special cases, but
3368 almost all the interesting cases are handled by creating
3369 a LAMBDA expression, calling COMPILE on it, then calling
3370 FUNCALL on the result.
3371 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3372 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3373 Since the code is newer, there might still be some new bugs
3374 (though not as many as before Martin Atzmueller's fixes:-). But
3375 the new code is substantially simpler and clearer, and hopefully
3376 any remaining bugs will be simpler, less fundamental, and more
3377 fixable then the bugs in the old code.
3378 ** The revised compiler is still a little unsteady on its feet.
3380 *** The debugging information it produces (particularly the names
3381 of FUNCTION objects) is sometimes much less useful than what
3382 the old compiler produced.
3383 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3384 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3385 within a MACROLET) has been temporarily weakened.
3386 ** There are new compiler optimizations for various functions:
3387 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3388 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3389 *** the math functions TRUNCATE, FLOOR, and CEILING
3390 *** the function-of-all-trades COERCE
3391 Mostly these should be transparent, but there's one
3392 potentially-annoying problem (bug 117): when the compiler
3393 inline-expands a function and does type analysis on the result,
3394 it can create control paths which have type mismatches, and
3395 when it can't prove that those control paths aren't taken,
3396 it will issue WARNINGs about the type mismatches. This is
3397 a particular problem in practice for the new sequence functions.
3398 It's not clear how this should be fixed, and for now, a
3399 workaround is given in the entry for 117 in the BUGS file.
3400 ** (Because of the interaction between the two previous items --
3401 occasional inlining problems and new inline expansions -- some
3402 of the new sequence function optimizations won't really kick in
3403 completely until debugging information, and then inlining, are
3404 straightened out in some future version.)
3405 * minor incompatible changes:
3406 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3407 DIRECTORY no longer implicitly promotes NIL slots of its
3408 pathname argument to :WILD. In particular, when you ask for the
3409 contents of a directory (which you used to be able to do without
3410 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3411 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3412 ** changes in behavior that ANSI explicitly defines to be
3413 implementation dependent:
3414 *** The new compiler-only implementation still conforms with ANSI,
3415 but acts a little different than before. Besides the obvious
3416 changes in performance tradeoffs (that the cost per form passed
3417 to EVAL has gone up, and the cost per form executed by EVAL
3418 has gone down), the behavior of the system changes a little
3419 because there are no longer any interpreted function objects.
3420 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3421 e.g. doing COMPILE on the output of interactive DEFUN is
3423 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3425 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3426 has changed from (:CL) to NIL.
3427 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3428 e.g. #\Newline and #\Tab, don't have names specified in the
3429 ANSI Common Lisp standard, is now based on their ASCII symbolic
3430 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3431 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3432 are no longer used for output.
3433 ** changes in internal implementation constants:
3434 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3435 4 million. (If your application spends a lot of time GCing and
3436 you have a lot of RAM, you might want to experiment with
3437 increasing it even more.)
3438 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3439 However, almost all old code should still continue to work without
3440 immediate update, as SB-C-CALL is now a (deprecated) nickname
3442 ** Old operator names in the style DEF-FOO are now deprecated in
3443 favor of new corresponding names DEFINE-FOO, for consistency with
3444 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3445 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3446 internal symbols, but a few supported extensions like
3447 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3448 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3449 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3450 etc. as you get deeper into recursive calls to the debugger
3451 command loop, instead of the old "5]", "5]]", "5]]]"
3452 sequence. (I was motivated to do this when squabbles between
3453 ILISP and SBCL left me very deeply nested in the debugger. In the
3454 short term, this change will probably provoke more ILISP/SBCL
3455 squabbles, but hopefully it will be an improvement in the long run.)
3456 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3457 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3458 function, since declaiming return types involving aliens is
3459 (1) annoyingly messy to do by hand and (2) vital to efficient
3460 compilation of code which calls such functions.
3461 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3462 longer reexported by the SB-EXT package. They're solely useful
3463 for alien code, so it seems more logical that you should get
3464 them from the SB-ALIEN package, not in SB-EXT.
3465 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3466 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3467 features. Instead, the code that they used to control is always
3468 built into the system.
3469 * many other bug fixes
3470 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3471 advantage of the new EVAL-WHEN stuff and to clean them up in
3472 general, and they are now more ANSI-compliant in a number of
3473 ways. Martin Atzmueller is responsible for a lot of this.
3474 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3476 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3477 *** correct ERROR type for various file operations
3478 *** some fixes for Lisp streams
3479 *** DEFMETHOD syntax checking
3480 *** changing old weird representation of debug information as
3481 strings (which, among their other deficiencies, don't transform
3482 correctly when you rename packages, and don't change their
3483 print representation when you change things like *PACKAGE*
3484 and *PRINT-LENGTH*) to symbols and lists of symbols
3485 He also made several improvements and fixed several bugs in DESCRIBE.
3486 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3488 *** misbehavior of WRITE-STRING/WRITE-LINE
3489 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3490 and several other LOOP problems as well
3491 *** DIRECTORY when similar filenames are present
3492 *** DEFGENERIC with :METHOD options
3493 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3494 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3495 *** argument ordering in FIND with :TEST option
3496 *** mishandled package designator argument in APROPOS-LIST
3497 *** various problems in the backquote readmacro
3498 *** a bug in APROPOS
3499 *** probably some others that I'm not describing very well here,
3500 since the CVS log documents them by reference to sbcl-devel
3501 messages, and the SourceForge archives aren't working well.:-(
3502 ** Dan Barlow improved the Alpha port (and is making progress on the
3503 PPC port, for those of you who think different).
3504 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3505 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3506 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3507 continues to work on the SPARC port (for those of you in a position
3508 to look down upon our little PC-compatible boxes from a great height).
3509 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3510 and uses that copy, rather than the current dynamic value, when
3511 it's trying to decide whether to truncate output. Thus e.g.
3512 (let ((*print-lines* 50))
3513 (pprint-logical-block (stream nil)
3515 (let ((*print-lines* 8))
3516 (print (aref possiblybigthings i) stream)))))
3517 should now truncate the logical block only at 50 lines, instead of
3518 often truncating it at 8 lines, as it did before.
3519 * The doc/cmucl/ directory, containing old CMU CL documentation
3520 from the time of the fork, is no longer part of the base system.
3521 SourceForge has shut down its anonymous FTP service, and with it
3522 my original plan for distributing the old CMU CL documentation
3523 there. For now, if you need these files you can download an old
3524 SBCL source release and extract them from it.
3525 * The fasl file version number changed again, for dozens of reasons,
3526 some of which are apparent above.
3528 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3529 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3530 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3531 patch, so that DISASSEMBLE output is much nicer.
3532 * The code in the SB-PROFILE package now seems reasonably stable.
3533 I still haven't decided what the final interface should look like
3534 (I'd like PROFILE to interact cleanly with TRACE, since both
3535 facilities use function encapsulation) but if you have a need
3536 for profiling now, you can probably use it successfully with
3537 the current CMU-CL-style interface.
3538 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3539 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3540 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3541 link now returns the dangling link itself, and for similar
3542 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3543 link itself. (In these cases the old code signalled an error and
3544 looped endlessly, respectively.) Thus, DIRECTORY now works even
3545 in the presence of dangling and cyclic symbolic links.
3546 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3547 is now a supported extension again, since the consensus on
3548 sbcl-devel was that it can be useful for ordinary development
3549 work, not just for debugging SBCL itself.
3550 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3551 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3552 that currently-defined functions might be redefined later with
3553 different return types.
3554 * Hash tables can be printed readably, as inspired by CMU CL code
3555 of Eric Marsden and SBCL code of Martin Atzmueller.
3556 * better error handling in CLOS method combination, thanks to
3557 Martin Atzmueller porting Pierre Mai's CMU CL patches
3558 * more overflow fixes for >16Mbyte I/O buffers
3559 * A bug in READ has been fixed, so that now a single Ctrl-D
3560 character suffices to cause end-of-file on character streams.
3561 In particular, now you only need one Ctrl-D at the command
3562 line (not two) to exit SBCL.
3563 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3565 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3566 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3567 does the right thing.
3568 * The implementation of some type tests, especially for CONDITION
3569 types, is now tidier and maybe faster, due to CMU CL code
3570 originally by Douglas Crosher, ported by Martin Atzmueller.
3571 * Some math functions have been fixed, and there are new
3572 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3573 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3574 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3575 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3576 likely to be built into sbcl-0.7.0.)
3577 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3578 CL, ported to SBCL by Martin Atzmueller)
3579 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3580 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3581 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3582 * There's a new slam.sh hack to shorten the edit/compile/debug
3583 cycle for low-level changes to SBCL itself, and a new
3584 :SB-AFTER-XC-CORE target feature to control the generation of
3585 the after-xc.core file needed by slam.sh.
3586 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3587 COMPILE-FILE is no longer supported, so that now every function
3588 gets an entry point, so that block compilation looks a little
3589 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3590 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3591 returns the number of bytes consed since the system started,
3592 rather than the number consed since the first time the function
3593 was called. (The new definition parallels ANSI functions like
3594 CL:GET-INTERNAL-RUN-TIME.)
3595 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3596 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3597 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3598 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3599 * Fasl file version numbers are now independent of the target CPU,
3600 since historically most system changes which required version
3601 number changes have affected all CPUs equally. Similarly,
3602 the byte fasl file version is now equal to the ordinary
3605 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3606 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3607 is no longer recognized. I apologize for this, because it was
3608 listed in SB-EXT as a supported extension, but I found that
3609 its existing behavior was poorly specified, as well as incorrectly
3610 specified, and it looked like too much of a mess to straighten it
3611 out. I have enough on my hands trying to get ANSI stuff to work..
3612 * many patches ported from CMU CL by Martin Atzmueller, with
3613 half a dozen bug fixes in pretty-printing and the debugger, and
3614 half a dozen others elsewhere
3615 * fixed bug 13: Floating point infinities are now supported again.
3616 They might still be a little bit flaky, but thanks to bug reports
3617 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3618 as flaky as they were.
3619 * The --noprogrammer command line option is now supported. (Its
3620 behavior is slightly different in detail from what the old man
3621 page claimed it would do, but it's still appropriate under the
3622 same circumstances that the man page talks about.)
3623 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3624 are now supported, and enabled by default. Thus, the compiler can
3625 handle many floating point and complex operations much less
3626 inefficiently. (Thus e.g. you can implement a complex FFT
3628 * The compiler now detects type mismatches between DECLAIM FTYPE
3629 and DEFUN better, and implements CHECK-TYPE more correctly, and
3630 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3631 thanks to patches from Martin Atzmueller.
3632 * various fixes to make the cross-compiler more portable to
3633 ANSI-conforming-but-different cross-compilation hosts (notably
3634 Lispworks for Windows, following bug reports from Arthur Lemmens)
3635 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3636 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3637 fixed thanks to Pierre Mai's CMU CL patches.
3638 * fixes to keep the system from overflowing internal counters when
3639 it tries to use i/o buffers larger than 16M bytes
3640 * fixed bug 45a: Various internal functions required to support
3641 complex special functions have been merged from CMU CL sources.
3642 (When I was first setting up SBCL, I misunderstood a compile-time
3643 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3644 * improved support for type intersection and union, fixing bug 12
3645 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3646 more obscure bugs as well
3647 * some steps toward byte-compiling non-performance-critical
3648 parts of the system, courtesy of patches from Martin Atzmueller
3649 * Christophe Rhodes has made some debian packages of sbcl at
3650 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3651 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3652 stable, but are nonetheless usable. When he's ready, I'd be happy
3653 to add them to the SourceForge "File Releases" section. (And if
3654 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3655 * new fasl file format version number (because of changes in
3656 internal representation of (OR ..) types to accommodate the new
3657 support for (AND ..) types, among other things)
3659 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3660 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3662 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3664 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3665 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3666 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3667 * DESCRIBE now works on structure objects again.
3668 * Most function call argument type mismatches are now handled as
3669 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3670 know whether the function will be redefined before the call is
3671 executed. (The compiler could flag local calls with full WARNINGs,
3672 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3673 it doesn't keep track of enough information to know whether calls
3674 are local in this sense.)
3675 * Compiler output is now more verbose, with messages truncated
3676 later than before. (There should be some supported way for users
3677 to override the default verbosity, but I haven't decided how to
3678 provide it yet, so this behavior is still controlled by the internal
3679 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3680 src/compiler/ir1util.lisp.)
3681 * Fasl file format version numbers have increased again, because
3682 support for the Gray streams extension changes the layout of the
3683 system's STREAM objects.
3684 * The Gray subclassable streams extension now works, thanks to a
3685 patch from Martin Atzmueller.
3686 * The full LOAD-FOREIGN extension (not just the primitive
3687 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3688 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3689 but like most other programs, it defaults to copying the Unix
3690 environment from the original process instead of starting the
3691 new process in an empty environment.
3692 * Extensions which manipulate the Unix environment now support
3693 an :ENVIRONMENT keyword option which doesn't smash case or
3694 do other bad things. The CMU-CL-style :ENV option is retained
3695 for porting convenience.
3696 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3697 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3699 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3701 * A patch from Martin Atzmueller seems to have solved the SIGINT
3702 problem, and as far as we know, signal-handling now works cleanly.
3703 (If you find any new bugs, please report them!)
3704 * The system no longer defaults Lisp source file names to types
3705 ".l", ".cl", or ".lsp", but only to ".lisp".
3706 * The compiler no longer uses special default file extensions for
3707 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3708 seems to expect a single default extension for all compiled code,
3709 and there's no compelling reason to try to stretch the standard
3710 to allow two different extensions.) Instead, byte-compiled files
3711 default to the same extension as native-compiled files.
3712 * Fasl file format version numbers have increased again, because
3713 a rearrangement of internal implementation packages made some
3714 dumped symbols in old fasl files unreadable in new cores.
3715 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3716 many fewer weird special cases.
3717 * Bug #17 (differing COMPILE-FILE behavior between logical and
3718 physical pathnames) has been fixed, and some related misbehavior too,
3719 thanks to a patch from Martin Atzmueller.
3720 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3721 by Tim Moore, ported to SBCL by Martin Atzmueller.
3722 * Martin Atzmueller fixed several filesystem-related problems,
3723 including bug #36, in part by porting CMU CL patches, which were
3724 written in part by Paul Werkowski.
3725 * More compiler warnings in src/runtime/ are gone, thanks to
3726 more patches from Martin Atzmueller.
3727 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3730 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3732 * DESCRIBE now works on CONDITION objects.
3733 * The debugger now handles errors which arise when trying to print
3734 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3736 * The build system now uses an additional file, customize-target-features.lisp,
3737 to allow local modifications to the target *FEATURES* list. (The point of
3738 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3739 debugging features enabled, without having to worry about propagating it
3740 into everyone's system when I do a "cvs update".) When no
3741 customize-target-features.lisp file exists, the target *FEATURES* list
3742 should be constructed the same way as before.
3743 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3744 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3745 warn about behavior which is undefined under the ANSI spec. Note: This
3746 is specified by ANSI, but it's not very popular with programmers.
3747 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3748 macro in the SBCL sources for an example of a workaround which you
3749 might use to make portable ANSI-standard code which does what you want.
3750 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3751 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3752 when it's used as a non-toplevel form. (This is required in order
3753 to implement the DEFCONSTANT-EQX macro.)
3754 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3755 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3756 funny things, due to bugs in the implementation of EVAL-WHEN
3757 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3758 non-toplevel DEFCONSTANTs are uncommon.)
3759 * The core file version number and fasl file version number have been
3760 incremented, because the old noncompliant DEFCONSTANT behavior involved
3761 calling functions which no longer exist, and because I also took the
3762 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3763 * fixed bug 1 (error handling before read-eval-print loop starts), and
3764 redid debugger restarts and related debugger commands somewhat while
3766 ** The QUIT debugger command is gone, since it did something
3767 rather different than the SB-EXT:QUIT command, and since it never
3768 worked properly outside the main toplevel read/eval/print loop.
3769 Invoking the new TOPLEVEL restart provides the same functionality.
3770 ** The GO debugger command is also gone, since you can just invoke
3771 the CONTINUE restart directly instead.
3772 ** The TOP debugger command is also gone, since it's redundant with the
3773 FRAME 0 command, and since it interfered with abbreviations for the
3775 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3777 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3778 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3779 some porting work by Martin Atzmueller)
3780 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3781 with a boatload of patches which he ported to SBCL. Now that those
3783 ** The system tries to make sure that its low-priority messages
3784 are prefixed by semicolons, to help people who like to use
3785 syntax highlighting in their ILISP buffer. (This patch
3786 was originally due to Raymond Toy.)
3787 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3788 patches originally written by Raymond Toy.
3789 ** The compiler understands coercion between single-value and
3790 multiple-VALUES type expressions better, getting rid of some very
3791 weird behavior, thanks to patches originally by Robert MacLachlan
3792 and Douglas Crosher.
3793 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3794 lambda lists, thanks to a patch originally by Pierre Mai.
3795 ** The system no longer bogusly warns about "abbreviated type
3797 ** The compiler gets less confused by inlining and RETURN-FROM,
3798 thanks to some patches originally by Tim Moore.
3799 ** The system no longer hangs when dumping circular lists to fasl
3800 files, thanks to a patch originally from Douglas Crosher.
3801 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3802 undefined function error.
3803 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3804 in structure slot initforms) for now, documented workaround instead:-|
3805 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3806 * fixed bug 5: added stubs for various Gray stream functions called
3807 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3808 installed, at least appropriate type errors are generated
3809 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3810 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3811 consistently in DEFMETHOD forms.
3812 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3813 it doesn't seem to affect SBCL after all
3814 * The C runtime system now builds with better optimization and many
3815 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3817 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3819 * The system is now under CVS at SourceForge (instead of the
3820 CVS repository on my home machine).
3821 * The new signal handling code has been tweaked to treat register
3822 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3823 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3824 downstream, e.g. in debug-int.lisp, has implicit dependencies
3825 on the unsignedness of integer representation of machine words,
3826 and that caused the system to bomb out with infinite regress
3827 when trying to recover from type errors involving signed values,
3828 e.g. (BUTLAST '(1 2 3) -1).)
3829 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3830 (This was one of the bugs Peter Van Eynde reported back in July.)
3831 * The system now uses code inspired by Colin Walters' O(N)
3832 implementation of MAP (from the cmucl-imp@cons.org mailing
3833 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3834 inline the MAP operation, and there is more than one
3835 sequence argument to the MAP call (so that it can't just
3836 do ETYPECASE once and for all based on the type of the
3837 single sequence argument). (The old non-inline implementation
3838 of the general M-argument sequence-of-length-N case required
3839 O(M*N*N) time when any of the sequence arguments were LISTs.)
3840 * The QUIT :UNIX-CODE keyword argument has been renamed to
3841 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3842 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3844 * Raymond Wiker's patches to port dynamic loading from Linux to
3845 FreeBSD have been added.
3846 * The BUGS file is now more nearly up to date, thanks in large part
3847 to Martin Atzmueller's review of it.
3848 * The debugger now flushes standard output streams before it begins
3849 its output ("debugger invoked" and so forth).
3850 * The core version number and fasl file version number have both
3851 been incremented, because of incompatible changes in the layout
3853 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3854 used to not be called for a saved Lisp image.)
3855 * Martin Atzmueller's version of a patch to fix a compiler crash,
3856 as posted on sbcl-devel 13 September 2000, has been installed.
3857 * Instead of installing Martin Atzmueller's patch for the
3858 compiler transform for SUBSEQ, I deleted the compiler transform,
3859 and transforms for some similar consing operations.
3860 * A bug in signal handling which kept TRACE from working on OpenBSD
3862 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3863 away by constant folding
3864 * The system now defines its address space constants in one place
3865 (in the Lisp sources), and propagates them automatically elsewhere
3866 (through GENESIS and the sbcl.h file). Therefore, patching the
3867 address map is less unnecessarily tedious and error-prone. The
3868 Lisp names of address space constants have also been systematized.
3869 * CVS tags like dollar-Header-dollar have been removed from
3870 the sources, because they have never saved me trouble and
3871 they've been source of trouble working with patches and other
3872 diff-related operations.
3873 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3874 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3876 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3878 * The system has been ported to OpenBSD.
3879 * The system now compiles with a simple "sh make.sh" on the systems
3880 that it's supported on. I.e., now you no longer need to tweak
3881 text in the target-features.lisp-expr and symlinks in src/runtime/
3882 by hand, the make.sh takes care of it for you.
3883 * The system is no longer so grossly inefficient when compiling code
3884 involving vectors implemented as general (not simple) vectors (VECTOR T),
3885 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3886 which dares to use the various sequence functions on non-simple
3887 vectors, takes less of a performance hit.
3888 * There is now a primitive type predicate VECTOR-T-P
3889 to test for the (VECTOR T) type, so that e.g.
3890 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3891 can now be compiled with some semblance of efficiency. (The old code
3892 turned the type declaration into a full call to %TYPEP at runtime!)
3893 * AREF on (VECTOR T) is still not fast, since it's still compiled
3894 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3895 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3896 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3897 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3898 without making full calls to SUBTYPEP at runtime.
3899 (Some analogous efficiency issues for non-simple vectors specialized to
3900 element types other than T, or for non-simple multidimensional arrays,
3901 have not been addressed. They could almost certainly be handled the
3902 same way if anyone is motivated to do so.)
3903 * The changes in array handling break binary compatibility, so
3904 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3905 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3906 instead of (VALUES T T).
3907 * By following the instructions that Dan Barlow posted to sbcl-devel
3908 on 2 July 2000, I was able to enable primitive dynamic object
3909 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3910 functionality is not implemented (since it calls ld to resolve
3911 library references automatically, requiring RUN-PROGRAM for its
3912 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3913 try to resolve library references) is now supported.
3914 * The system now flushes the standard output streams when it terminates,
3915 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3916 them at several other probably-convenient times, e.g. in each pass of
3917 the toplevel read-eval-print loop, and after evaluating a form given
3918 as an "--eval" command-line option. (These changes were motivated by a
3919 discussion of stream flushing issues on cmucl-imp in August 2000.)
3920 * The source transform for TYPEP of array types no longer assumes
3921 that an array whose element type is a not-yet-defined type
3922 is implemented as an array of T, but instead punts, so that the
3923 type will be interpreted at runtime.
3924 * There is now some support for cross-compiling in make.sh: each of
3925 the phases of make.sh has its own script. (This should be transparent
3926 to people doing ordinary, non-cross-compile builds.)
3927 * Since my laptop doesn't have hundreds of megabytes of memory like
3928 my desktop machine, I became more motivated to do some items on
3929 my to-do list in order to reduce the size of the system a little:
3930 ** Arrange for various needed-only-at-cold-init things to be
3931 uninterned after cold init. To support this, those things have
3932 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3933 symbols with such names are now uninterned after cold init).
3934 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3935 fasl files for cold load.
3936 ** Remove the old compiler structure pooling code (which used to
3937 be conditional on the target feature :SB-ALLOC) completely.
3938 ** Redo the representation of some data in cold init to be more compact.
3939 (I also looked into supporting byte compiled code at bootstrap time,
3940 which would probably reduce the size of the system a lot, but that
3941 looked too complicated, so I punted for now.)
3942 * The maximum signal nesting depth in the src/runtime/ support code has
3943 been reduced from 4096 to 256. (I don't know any reason for the very
3944 large old value. If the new smaller value turns out to break something,
3945 I'll probably just bump it back up.)
3946 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3948 * Many, many bugs reported by Peter Van Eynde have been added to
3949 the BUGS list; some have even been fixed.
3950 * While enabling dynamic object file loading, I tried to make the
3951 code easier to understand, renaming various functions and variables
3952 with less ambiguous names, and changing some function calling
3953 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3954 * While trying to figure out how to do the OpenBSD port, I tried to
3955 clean up some of the code in src/runtime/. In particular, I dropped
3956 support for non-POSIX signal handling, added various comments,
3957 tweaked the code to reduce the number of compilation warnings, and
3958 renamed some files to increase consistency.
3959 * To support the new automatic configuration functionality in make.sh,
3960 the source file target-features.lisp-expr has been replaced with the
3961 source file base-target-features.lisp-expr and the machine-generated
3962 file local-target-features.lisp-expr.
3963 * fixed a stupid quoting error in make.sh so that using CMU CL
3964 "lisp -batch" as cross-compilation host works again
3966 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3968 * DESCRIBE no longer tries to call itself recursively to describe
3969 bound/fbound values, so that it no longer fails on symbols which are
3970 bound to themselves (like keywords, T, and NIL).
3971 * DESCRIBE now works on generic functions.
3972 * The printer now prints less-screwed-up representations of closures
3973 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3974 * A private symbol is used instead of the :EMPTY keyword previously
3975 used to mark empty slots in hash tables. Thus
3976 (DEFVAR *HT* (MAKE-HASH-TABLE))
3977 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3978 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3979 now does what ANSI says that it should. (You can still get
3980 similar noncompliant behavior if bang on the hash table
3981 implementation with all the symbols you get back from
3982 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3983 This breaks binary compatibility, since tests for equality to
3984 :EMPTY are wired into things like the macroexpansion of
3985 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3987 * There's now a minimal placeholder implementation for CL:STEP,
3988 as required by ANSI.
3989 * An obscure bug in the interaction of the normal compiler, the byte
3990 compiler, inlining, and structure predicates has been patched
3991 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3992 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3994 * Missing ordinary arguments in a macro call are now detected even
3995 when the macro lambda list contains &KEY or &REST.
3996 * The debugger no longer complains about encountering the top of the
3997 stack when you type "FRAME 0" to explicitly instruct it to go to
3998 the top of the stack. And it now prints the frame you request even
3999 if it's the current frame (instead of saying "You are here.").
4000 * As specified by ANSI, the system now always prints keywords
4001 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4003 * The default initial SIZE of HASH-TABLEs is now smaller.
4004 * Type information from CLOS class dispatch is now propagated
4005 into DEFMETHOD bodies, so that e.g.
4006 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4008 is now basically equivalent to
4009 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4010 (DECLARE (TYPE SINGLE-FLOAT X))
4012 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4013 operation, without having to do run-time type dispatch.
4014 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4015 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4016 * Since I don't seem to be making big file reorganizations very often
4017 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4018 is overflowing my ability to conveniently back them up), I've finally
4019 checked the system into CVS. (The CVS repository is on my home system,
4020 not at SourceForge -- putting it on SourceForge might come later.)
4021 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4022 high-level GC-NOTIFY-FOO functions send their output. (There's
4023 still very little control of where low-level verbose GC functions
4024 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4025 less than it used to -- the GC-NOTIFY-FOO functions are now under
4026 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4027 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4028 in only one place in the source code, and propagates it automatically
4029 everywhere that it's needed. Thus e.g. when I bump the version from
4030 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4031 * The C source files now include boilerplate legalese and documentation
4032 at the head of each file (just as the Lisp source files already did).
4033 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4034 to his page will be replaced with a link to his new CLiki service.
4036 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4038 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4039 * The build process now looks for GNU make under the default name "gmake",
4040 instead of "make" as it used to. If GNU make is not available as "gmake"
4041 on your system, you can change this default behavior by setting the
4042 GNUMAKE environment variable.
4043 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4044 can build without error under CMU CL.
4046 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4048 * There is now a partial SBCL user manual (with some new text and some
4049 text cribbed from the CMU CL manual).
4050 * The beginnings of a profiler have been added (starting with the
4051 CMU CL profiler and simplifying and cleaning up). Eventually the
4052 main interface should be through the TRACE macro, but for now,
4053 it's still accessed through vaguely CMU-CL-style functions and macros
4054 exported from the package SB-PROFILE.
4055 * Some problems left over from porting CMU CL to the new
4056 cross-compilation bootstrap process have been cleaned up:
4057 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4058 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4059 part when getting rid of the compiler *BACKEND* stuff.)
4060 ** The value of *NULL-TYPE* was screwed up, because it was
4061 being initialized before the type system knew the final
4062 definition of the 'NULL type. This screwed up several key
4063 optimizations in the compiler, causing inefficiency in all sorts
4064 of places. (I found it because I wanted to understand why
4065 GET-INTERNAL-RUN-TIME was consing.)
4066 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4067 PROCLAIM FTYPE information. Unfortunately this broke binary
4068 compatibility again, since now the forms output by DEFGENERIC
4069 to refer to functions which didn't exist in 0.6.3.
4070 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4071 can use the new (as of 0.6.3) transform for SOME into MAP into
4073 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4074 values to (INTEGER 0 1000000), so that the time code will no longer
4075 occasionally get blown up by Linux returning 1000000 microseconds
4076 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4077 its output conform to the ANSI spec. (Alas, this makes its output
4078 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4079 * A full call to MAP NIL with a single sequence argument no longer conses.
4080 * fixes to problems pointed out by Martin Atzmueller:
4081 * The manual page no longer talks about multiprocessing as though
4082 it were currently supported.
4083 * The ILISP support patches have been removed from the distribution,
4084 because as of version 5.10.1, ILISP now supports SBCL without us
4085 having to maintain patches.
4086 * added a modified version of Raymond Toy's recent CMU CL patch for
4087 EQUALP comparison of HASH-TABLE
4089 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4091 * The system still can't cross-compile itself with
4092 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4093 entails), but at least it can compile more of itself that way
4094 than it used to be able to, and various buglets which were uncovered
4095 by trying to cross-compile itself that way have now been fixed.
4096 * This release breaks binary compatibility again. This time
4097 at least I've incremented the FASL file format version to 2, so that the
4098 problem can be detected reliably instead of just causing weird errors.
4099 * various new style warnings:
4100 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4101 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4102 ** using the lexical binding of a variable named in the *FOO* style
4103 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4104 as specified by ANSI.
4105 * *RANDOM-STATE* is no longer automatically initialized from
4106 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4107 default behavior of the system is to repeat its behavior every time
4108 it's run. If you'd like to change this behavior, you can always
4109 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4110 old convention there was no comparably easy way to get the system to
4111 repeat its behavior every time it was run.
4112 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4113 FTYPE declarations has been removed, in favor of their ANSI
4114 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4115 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4116 the types of their sequence arguments can be determined at compile time.
4117 This is done through a new open code expansion for MAP which eliminates
4118 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4119 argument types can be determined at compile time.
4120 * The optimizer now transforms COERCE into an identity operation when it
4121 can prove that the coerced object is already of the correct type. (This
4122 can be a win for machine generated code, including the output of other
4123 optimization transforms, such as the MAP transform above.)
4124 * Credit information has been moved from source file headers into CREDITS.
4125 * Source file headers have been made more standard.
4126 * The CASE macro now compiles without complaining even when it has
4129 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4131 * (Note that the way that the PCL macroexpansions were rewritten
4132 to accommodate the change in DEFGENERIC below breaks binary
4133 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4134 not run under sbcl-0.6.2. Once we get out of alpha releases,
4135 i.e. hit release 1.0.0, we'll probably try to maintain binary
4136 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4137 and sbcl-1.4.4. Until then, however, it might be fairly common
4138 for maintenance releases to break binary compatibility.)
4139 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4141 * The reporting of unhandled signals has been changed to print some
4142 explanatory text as well as the report form. (Previously only
4143 the report form was printed.)
4144 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4145 it defines, so that the compiler no longer issues undefined function
4146 warnings for compiled-but-not-yet-loaded generic functions.
4147 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4148 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4149 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4150 section of the ANSI spec are supported.
4151 * The "" nickname for the "KEYWORD" package has been removed.
4152 The reader still handles symbol tokens which begin with a package marker
4153 as keywords, but it doesn't expose its mechanism for doing so in the
4154 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4155 * The system now issues STYLE-WARNINGs for contradictory TYPE
4156 proclamations. (Warnings for contradictory FTYPE proclamations would
4157 be nice too, but those can't be done usefully unless the type system
4158 is made smarter about FUNCTION types.)
4159 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4160 systematized, so that "*target-*.lisp is supposed to exist only on the
4161 target and imply that there's a related file which exists on the
4162 host, and *host-*.lisp is supposed to exist only on the host and imply
4163 that there's a related file which exists on the target. This involves a
4164 lot of renaming. Hopefully the acute confusion caused by the renaming
4165 will be justified by the reduction in chronic confusion..
4166 ** runtime-type.lisp -> early-target-type.lisp
4167 ** target-type.lisp -> late-target-type.lisp
4168 ** early-host-format.lisp -> early-format.lisp
4169 ** late-host-format.lisp -> late-format.lisp
4170 ** host-error.lisp -> misc-error.lisp
4171 ** early-error.lisp -> early-target-error.lisp
4172 ** late-error.lisp -> late-target-error.lisp
4173 ** host-defboot.lisp -> early-defboot.lisp
4174 ** code/misc.lisp -> code/target-misc.lisp
4175 ** code/host-misc.lisp -> code/misc.lisp
4176 ** code/numbers.lisp -> code/target-numbers.lisp
4177 ** code/early-numbers.lisp -> numbers.lisp
4178 ** early-host-type.lisp -> early-type.lisp
4179 ** late-host-type.lisp -> late-type.lisp
4180 ** host-typep.lisp -> typep.lisp
4181 ** load.lisp -> target-load.lisp
4182 ** host-load.lisp -> load.lisp
4183 ** host-disassem.lisp -> disassem.lisp
4184 ** host-insts.lisp -> insts.lisp
4185 ** byte-comp.lisp -> target-byte-comp.lisp
4186 ** host-byte-comp.lisp -> byte-comp.lisp
4187 ** host-signal.lisp -> signal.lisp
4188 ** host-defstruct.lisp -> defstruct.lisp
4189 ** late-target-type.lisp -> deftypes-for-target.lisp
4190 Furthermore, several other previously target-only files foo.lisp (e.g.
4191 hash-table.lisp and random.lisp) have been split into a target-and-host
4192 foo.lisp file and a target-only target-foo.lisp file, with their key type
4193 definitions in the target-and-host part, so that the cross-compiler will
4194 know more about target types.
4195 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4196 gone away. In their place are various *BACKEND-FOO* variables
4197 corresponding to the slots of the old structure.
4198 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4199 into the target SBCL has been fixed.
4200 * The chill.lisp system for loading cold code into a running SBCL
4202 * Support for the CMU CL "scavenger hook" extension has been removed.
4203 (It was undocumented and unused in the CMU CL sources that SBCL was
4204 derived from, and stale in sbcl-0.6.1.)
4205 * Various errors in the cross-compiler type system were detected
4206 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4207 (enabling various consistency checks). Many of them were fixed,
4208 but some hard problems remain, so the compiler is back to
4209 running without *TYPE-SYSTEM-INITIALIZED* for now.
4210 * As part of the cross-compiler type system cleanup, I implemented
4211 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4212 * I have started adding UNCROSS calls throughout the type system
4213 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4214 on cross-compiler input files will be able to go away, and various
4216 * CONSTANTP now returns true for quoted forms (as explicitly required
4219 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4221 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4222 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4223 (MAKE-LIST -1). In the longer term, it ought to become true
4224 that declarations are assertions even at SAFETY 1. For now, it's not
4225 quite true even at SAFETY 3, but it's at least more nearly true..
4226 (Note that this change seems to increases the size of the system by
4227 O(5%) and to decrease the speed of the compiler by 20% or more.)
4228 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4229 for the problem of printing dozens of lines of distracting information
4230 about low-level system machinery as part of the top stack frame
4231 on entry to the debugger when an undefined function was called.
4232 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4233 is not reset to COMMON-LISP-USER.
4234 * Compilation of stuff related to dyncount.lisp has been made conditional
4235 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4236 smaller. The various dyncount-related symbols have been moved into
4237 a new "SB-DYNCOUNT" package.
4238 * tty-inspect.lisp has been renamed to inspect.lisp.
4239 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4240 feature has gone away. (When we eventually port to other flavors of
4241 libc and/or Unix, we'll try to make the differences between flavors
4242 invisible at the user level.)
4243 * Various other *FEATURES* tags, and/or their associated conditionals,
4244 have been removed if obsolescent, or given better documentation, or
4245 sometimes given more-mnemonic names.
4247 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4249 * tidied up "make.sh" script
4250 * tidied up system directory structure
4251 * better "clean.sh" behavior
4252 * added doc/FOR-CMUCL-DEVELOPERS
4253 * many many small tweaks to output format, e.g. removing possibly-confusing
4254 trailing #\. character in DESCRIBE-INSTANCE
4255 * (EQUALP #\A 'A) no longer signals an error.
4256 * new hashing code, including EQUALP hashing
4257 * tidied up Lisp initialization and toplevel
4258 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4259 * command line argument processing
4260 * added POSIX-GETENV function to deal with Unix-ish environment variables
4261 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4262 terminating SBCL on EOF
4263 * non-verbose GC by default
4264 * There is no more "sbcl" shell script; the sbcl file is now the C
4265 runtime executable (just like CMU CL).
4266 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4268 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4269 symbol and package information as Lisp native symbol and package objects
4270 instead of strings naming symbols and strings naming packages. This way,
4271 whenever packages are renamed (as in warm init), debug information is
4272 transformed along with everything else.
4273 * tweaked the optimization policy declarations which control the building
4274 of SBCL itself. Now, among other things, the system no longer saves
4275 source location debugging information. (This helps two problems at once
4276 by reducing SBCL size and by keeping SBCL from trying to look for its
4277 sources -- which may not exist -- when reporting errors.)
4278 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4279 debugging and testing purposes
4280 * cleaned up printing, making the printer call PRINT-OBJECT for
4281 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4282 methods, giving nearly-ANSI behavior
4283 * converted almost all special variables to use *FOO* naming convention
4284 * deleted PARSE-TIME functionality, since it can be done portably
4285 * moved some files out of cold init into warm init
4286 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4288 * regularized formatting of source files
4289 * added an install.sh script
4290 * fixed ridiculous memory usage of cross-compiler by making
4291 compiler/alloc.lisp not try to do pooling unless it can hook
4292 itself into the GC of the cross-compilation host. Now the system
4293 builds nicely on my old laptop.
4294 * added :SB-ALLOC in target-features.lisp-expr
4295 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4296 was not implemented)
4297 * re-did condition handling and note reporting in the compiler. Notes
4298 are no longer handled by signalling conditions. Style warnings
4299 and warnings are handled more correctly and reported in such a way
4300 that it's easy to find one or the other in your output (so that you
4301 can e.g. figure out which of many problems caused COMPILE-FILE to
4303 * changed the severity of several compiler warnings from full WARNING
4304 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4305 compiler note reporting so that it doesn't use the condition system
4306 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4307 COMPILE-FILE command)
4308 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4309 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4310 no longer supported, since their implementation was deeply intertwingled
4311 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4312 * removed bogus "support" for compiler macros named (SETF FOO), and
4313 removed the compiler macro for SETF INFO (but only after making a fool
4314 of myself on the cmucl-imp mailing list by posting a bogus patch for
4315 DEFINE-COMPILER-MACRO..)
4316 * Compiled files containing forms which have side effects on the Lisp
4317 reader (such as DEFPACKAGE forms) are now handled more correctly.
4318 (Compiler queuing of top level lambdas has been suppressed by setting
4319 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4320 * deleted various currently-unused source files, e.g. gengc.lisp. They
4321 may be added back at some point e.g. when porting to other architectures,
4322 but until they are it's distracting to distribute them and to try to
4324 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4325 there were so many of them they're just distractions, and UNCROSS is
4326 known to be able to handle the current sources
4327 * moved PROFILE functionality into TRACE, so that it will be clear
4328 how the wrapping and unwrapping of functions when you profile them
4329 interacts with the wrapping and unwrapping of functions when you
4330 trace them. (Actually, the functionality isn't there yet, but at least
4331 the interface specification is there. Hopefully, the functionality will
4332 arrive with some future maintenance release.)
4333 * removed host-oops.lisp
4334 * changed signature of QUIT function to allow UNIX-CODE argument
4335 * fixed READ-SEQUENCE bug
4336 * tweaked verbose GC output so that it looks more like the progress
4337 output that ANSI specifies for functions like LOAD
4338 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4339 * added <http://sbcl.sourceforge.com> to the banner information printed by