1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
3 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
4 allows assining a global minimum value to optimization qualities
5 (overriding proclamations and declarations).
6 * enhancement: closed over variables can be stack-allocated on x86
8 * performance bug fix: GETHASH and (SETF GETHASH) are once again
10 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
11 * bug fix: using obsoleted structure instances with TYPEP and
12 generic functions now signals a sensible error.
13 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
14 (reported by Kristoffer Kvello)
15 * bug fix: backtrace construction is now more careful when making
16 lisp-objects from pointers on the stack, to avoid creating bogus
17 objects that can be seen by the GC.
18 * bug fix: defaulting of values in contexts expecting more than 7
19 variables now works on x86-64. (reported by Christopher Laux)
20 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
23 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
24 * MOP improvement: support for user-defined subclasses of
25 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
26 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
27 the expansion of DEFMETHOD, is responsible for generating a form
28 which creates a list of specializers when evaluated. Additional
29 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
30 debugging and introspective support.
31 * minor incompatible change: the (unsupported) spinlock interface
32 has changed: free spinlock now has the value NIL, and a held spinlock
33 has the owning thread as its value.
34 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
35 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
37 * enhancement: name of a socket-stream is now "a socket" instead of
39 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
40 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
41 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
42 for generic functions. (Thanks to Tobias C. Rittweiler)
43 * optimization: bignum printing speed has been improved by 20-40%
44 (depending on the bignum size.)
45 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
47 * bug fix: the cache used by the CLOS to store precomputed effective
48 methods, slot offsets, and constant return values is now thread and
50 * bug fix: generic function dispatch function updating is now thread
51 and interrupt safe (in the sense that the known issues have been
53 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
54 * bug fix: interning EQL-specializers is now thread and interrupt safe.
55 * bug fix: asdf systems with dependencies to the SB-POSIX or
56 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
57 * bug fix: interrupt safety of applicable method computation has been
60 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
61 * new contrib: sb-cover, an experimental code coverage tool, is included
63 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
65 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
66 produce more efficient code on x86-64 if the compiler can determine
67 that (+ INDEX <constant>) does not require a bounds check and FOO
68 has an element type at least 8 bits wide.
69 * enhancement: a new, experimental synchronous timeout facility is
70 provided. Refer to SB-SYS:WITH-DEADLINE for details.
71 * enhancement: when a symbol name conflict error arises, the
72 conflicting symbols are always printed with a package prefix.
73 (thanks to Kevin Reid)
74 * enhancement: stepping is now once again supported on the SPARC and
75 MIPS platforms. (It is also now more likely to work on CheneyGC
77 * enhancement: sb-sprof can now also track and report accurate call
79 * bug fixes: the treatment of non-standard subclasses of
80 SB-MOP:SPECIALIZER is more correct.
81 * incompatible change: PURIFY no longer copies the data from the
82 dynamic space into the static and read-only spaces on platforms
83 that use the generational garbage collector
84 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
86 * bug fix: GC race condition occasionally resulting in crashes with
87 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
89 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
91 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
92 a STYLE-WARNING to be signalled (regression from 1.0.4.)
93 * bug fix: an asynchronous interrupt could previously leave the
94 system running with GC inhibited.
95 * bug fix: a DECLARE form evaluated at top-level now causes an error
96 rather than silently (or verbosely) returning NIL.
97 * bug fix: trying to dissassemble functions compiled at high DEBUG could
98 result in a error being signalled due to source form lookup errors.
99 (reported by Peter Graves)
101 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
102 * incompatible change: removed writer methods for host-ent-name,
103 host-ent-addresses -- changing the values did not update the DNS
105 * minor incompatible change: changed experimental JOIN-THREAD interface
106 * documentation: the manual now lists reader and writer methods
107 in class slot documentation sections. (thanks to Richard M Kreuter)
108 * documentation: unwinding from asyncronous events has been
109 documented as unsafe.
110 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
111 in multithreaded application code.
112 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
113 platforms. (thanks to James Anderson for the optimization hint)
114 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
116 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
117 variants no longer cons.
118 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
119 their NOT- variants no longer cons.
120 * optimization: EQUAL hash tables no longer use SXHASH for objects
121 of all data types, but instead use an EQL hash for types for which
122 EQUAL is the same as EQL.
123 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
124 are significantly faster.
125 * optimization: new STRING-OUTPUT-STREAM implementation is much
126 faster for large outputs and conses 30% less on average.
127 * enhancement: XREF information is now collected for references made
128 to global variables using SYMBOL-VALUE and a constant argument.
129 * enhancement: SIGINT now causes a specific condition
130 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
131 * bug fix: FILE-POSITION used to signal an error in some cases where
132 ANSI requires it to return NIL.
133 * bug fix: ADJUST-ARRAY is now interrupt-safe.
134 * bug fix: adding and removing fd-handlers is now interrupt-safe.
135 * bug fix: inlined calls to C now ensure 16byte stack alignment on
137 * bug fix: bad type declaration in the CLOS implementation has
138 been fixed. (reported by James Anderson)
139 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
140 (thanks to Lutz Euler)
141 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
142 required by the CLHS. (reported by Samium Gromoff)
143 * bug fix: dead unbound variable references now signal an error.
144 (reported by Marco Monteiro)
145 * bug fix: / with an unused value was being deleted in safe code.
146 (thanks to Marco Monteiro and Kevin Reid)
147 * bug fix: number of characters that can be written onto a single
148 line in a file is unlimited.
149 * bug fix: some GC deadlocks caused by asynchronous interrupts have
150 been fixed by inhibiting interrupts for when GC is disbled.
151 * bug fix: some interrupt safety issues with GETHASH, (SETF
152 GETHASH), CLRHASH and REMHASH have been fixed.
153 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
154 specifier no longer causes infinite recursion.
155 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
156 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
157 * bug fix: an &environment argument with the correct variable information
158 is passed to macros that are expanded during byte compilation.
159 (reported by Samium Gromoff)
160 * bug fix: a logic error was causing the hash function for bignums to
161 have excessive amounts of collisions. (reported by Faré Rideau)
162 * bug fix: modifying the contents of an array could change the return
163 value of SXHASH on that array, which is only allowed for strings
164 and bit vectors. (bug introduced in 0.9.16)
165 * improvement: the x86-64/darwin port now passes all tests (except
166 for the debugger tests) but should still be considered
167 experimental until this is fixed.
168 * improvement: a style-warning is signaled for CASE (etc) clauses with
169 duplicate keys. (thanks to Kevin Reid)
170 * improvement: macos/x86 and macos/x86-64 now use mach exception
171 handlers for dealing with illegal instructions (for trapping and
172 error handling) and memory protection violations (for GC).
174 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
175 * new platform: experimental support for x86-64/darwin (MacOS).
176 * incompatible change: the thread-safe (on most platforms) getaddrinfo
177 and getnameinfo sockets functions are used instead of gethostbyaddr
178 and gethostbyname, on platforms where the newer functions are available.
179 As a result, the ALIASES field of HOST-ENT will always be NIL on these
181 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
182 don't print the startup banner, but behave as if --noinform was passed
183 as a command line argument. (thanks to Kevin Reid)
184 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
185 * optimization: code using alien values with undeclared types is much faster.
186 * optimization: the compiler is now able to open code SEARCH in more cases.
187 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
188 * bug fix: using standardized COMMON-LISP special variables as loop
189 variables no longer signals bogus package lock violations. (reported
191 * bug fix: declaring local loop variables to be of a range-limited type
192 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
193 (reported by Andras Simon)
194 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
195 bugs remain on x86-64.)
196 * bug fix: the #= and ## reader macros now interact reasonably with
197 funcallable instances.
198 * bug fix: type-checks for function arguments were compiled using the
199 compiler policy settings of the wrong lexical environment (in 1.0.2
201 * bug fix: SHADOW accepts characters as string designators, as required
202 by the spec (thanks to Eric Marsden)
203 * bug fix: fixed GC safety issues when foreign functions are called with
204 non-base strings as arguments
205 * bug fix: more consistent error output for fatal-compiler-warnings, like
207 * bug fix: functions with proper names were showing up as NIL in
208 backtraces (reported by Edi Weitz, regression in 1.0.2)
210 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
211 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
212 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
213 produce more efficient code on the x86 if the compiler can determine
214 that (+ INDEX <constant>) does not require a bounds check and FOO
215 has an element type at least 8 bits wide.
216 * bug fix: references to undefined variables are handled the same way
217 in toplevel forms as in the normal compiler
218 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
220 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
221 (thanks to Jon Buller)
222 * improvement: faster compilation times for complex functions
223 * improvement: added readlink support to SB-POSIX (thanks to Richard
226 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
227 * improvement: experimental support for mach exception handling on
228 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
230 * improvement: support for GBK external format.
231 (thanks to Chun Tian (binghe))
232 * improvement: the debugger now displays variables that have been closed
233 over, in code compiled with (DEBUG 2) or higher
234 * improvement: support for executable cores on NetBSD (thanks to
236 * new feature: added a RESTART-FRAME debugger command
237 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
238 be used to provide an implementation for FILE-POSITION on Gray streams
239 (thanks to Eric Marsden)
240 * new feature: add syslog, openlog and closelog support to SB-POSIX
241 (thanks to Richard Kreuter)
242 * optimization: the function call overhead in code compiled with
243 a high DEBUG optimization setting is significantly
244 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
245 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
246 (thanks to Tony Martinez)
247 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
248 evaluated once (reported by Kevin Reid)
249 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
250 works again (thanks to Kilian Sprotte)
251 * bug fix: an error is signaled for tagbodies with duplicate tags
252 (thanks to Stephen Wilson)
253 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
254 * bug fix: Win32 port can now handle foreign code unwinding Lisp
255 stack frames from alien callbacks.
256 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
257 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
258 65k pages (thanks to David Woodhouse)
259 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
261 changes in sbcl-1.0.1 relative to sbcl-1.0:
262 * new platform: FreeBSD/x86-64, including support for threading.
263 * new feature: the compiler stores cross-referencing information
264 abount function calls (who-calls), macroexpansion (who-macroexpands)
265 and special variables (who-binds, who-sets, who-references) for code
266 compiled with (< SPACE 3). This information is available through the
267 sb-introspect contrib.
268 * new feature: users may subclass SEQUENCE, and have instances of
269 these classes interoperate with standard Common Lisp functions if
270 a number of methods are defined. (This feature is experimental
271 and the interface subject to change based on feedback from SBCL
272 users and the general community)
273 * improvement: sb-sprof traces call stacks to an arbitrary depth on
274 x86/x86-64, rather than the previous fixed depth of 8
275 * improvement: another pthread back-end of mutex "pthread-futex".
276 * bug fix: non-ascii command-line arguments are processed correctly
277 (thanks to Yaroslav Kavenchuk)
278 * bug fix: non-required arguments were not passed correctly when a method
279 defined using DEFMETHOD was called from a mop-generated method using
280 CALL-NEXT-METHOD (reported by Pascal Costanza)
281 * bug fix: recursion is now permitted in accessors through
282 SLOT-UNBOUND. (reported by Pascal Costanza)
283 * bug fix: an error was signaled at startup if the HOME environment
284 variable was defined, but had an empty value (reported by Peter Van Eynde)
285 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
286 signal a PROGRAM-ERROR, not a BUG.
287 * bug fix: SB-POSIX:READDIR works when built with large file support.
288 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
290 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
291 for better automatic indentation support. (thanks to Matt Pillsbury)
292 * optimization: loading generic functions no longer takes O(n^2) time,
293 proportional to the amount of methods in the generic function
294 (reported by Todd Sabin and Jeremy Brown)
295 * optimization: the FIND and POSITION family of sequence functions
296 are significantly faster on arrays whose element types have been
298 * improvements to the Windows port:
299 ** Intermittent heap corruption problems have been fixed. (thanks
300 to Alastair Bridgewater)
301 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
303 ** Lisp is able to unwind foreign exception frames from alien
304 callbacks. (thanks to Alastair Bridgewater)
306 changes in sbcl-1.0 relative to sbcl-0.9.18:
307 * improvement: experimental support for threading on FreeBSD/x86.
308 (thanks to NIIMI Satoshi)
309 * improvement: runtime option --dynamic-space-size can be used
310 to set the size of the dynamic space reserved on startup.
311 * improvement: floating point modes in effect are now saved in
312 core, and restored on startup.
313 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
314 startup, not time since first call to GET-INTERNAL-REAL-TIME.
315 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
316 threads are not running after *SAVE-HOOKS* have run.
317 * improvement: writes to CLOS instance slots are type-checked in code
318 compiled with (SAFETY 3)
319 * improvement: floating-point exception handling on FreeBSD (thanks to
321 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
322 (thanks to Zach Beane)
323 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
325 * improvement: added support for the Shift-JIS external format.
326 (contributed by NIIMI Satoshi)
327 * improvement: callbacks are supported on Linux/PPC. (thanks to
329 * bug fix: compiler bug triggered by a (non-standard) VALUES
330 declaration in a LET* was fixed. (reported by Kaersten Poeck)
331 * bug fix: file compiler no longer confuses validated and already
332 dumped structurres. (reported by Kaersten Poeck)
333 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
334 fill-pointer signals a type-error as required. (thanks to
336 * bug fix: disassemly of funcallable instances works.
337 * bug fix: single stepping on PPC.
338 * bug fix: fix thread-safety problems in the type system (generally
339 manifesting as nonsensical errors like "STRING is a bad type specifier
340 for sequences" or "The value 1 is not of type FIXNUM")
341 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
342 (reported by Josip Gracin)
343 * bug fix: an error is signaled for attempts to displace arrays with
344 incompatible element types (thanks to Mario Mommer)
345 * bug fix: more correct handling of wide characters in debug info
346 (bug reported by Attila Lendvai and fixed by Juho Snellman)
347 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
348 and don't cause extra consing
349 * optimization: MAP and MAP-INTO are significantly faster on vectors
350 whose elements types have been declared.
351 * Improvements to SB-SPROF:
352 ** Support for allocation profiling
353 ** Reduced profiling overhead, especially for long profiling runs
354 * Improvements to the Windows port:
355 ** floating point exceptions are now reported correctly.
356 ** stack exhaustion detection works partially.
357 ** more accurate GET-INTERNAL-REAL-TIME.
358 ** better RUN-PROGRAM behaviour in case of errors in the child
360 ** PROBE-FILE now simplifies pathnames correctly.
361 ** DIRECTORY now works correctly with :WILD-INFERIORS.
362 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
363 1st 1970 (although time timezone data will be unreliable).
364 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
366 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
367 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
368 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
370 * bug fix: two potential GC deadlocks affecting threaded builds.
371 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
372 returns T (reported by Anton Kazennikov)
373 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
374 with non-variable places
375 * bug fix: remove a race condition in the setting of
376 funcallable-instance functions, this should make threaded CLOS
377 code more stable against memory faults.
378 * bug fix: corruption of specials when unbinding is interrupted by an
379 asynchronous unwind (reported by Hannu Koivisto)
380 * improvement: the debugger will now also display local variables that
381 are only used once, for code compiled with a DEBUG optimization quality
384 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
385 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
386 * incompatible change: External-format support for FFI calls. The
387 SB-ALIEN:C-STRING no longer implies an ASCII
388 external-format. Instead the string is subject to external-format
389 conversion. Additionally return values of type SB-ALIEN:C-STRING
390 are by default of type (SIMPLE-ARRAY CHARACTER), not
391 SIMPLE-BASE-STRING. For an alien type that behaves like the old
392 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
393 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
394 * incompatible change: SB-EXT package no longer contains the
395 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
396 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
398 * incompatible change: the single-stepper is no longer available
399 on Alpha, Mips and Sparc platforms.
400 * minor incompatible change: the direct superclasses of
401 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
402 not (STANDARD-OBJECT FUNCTION). This makes the
403 class-precedence-lists of GENERIC-FUNCTION and
404 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
406 * minor incompatible change: the default stream external format on
407 non-unicode SBCL is no longer detected from the locale, but is always
409 * new feature: Add a version of evaluator that uses an interpreter instead
410 of the compiler. EVAL still uses the compiler by default, to switch it
411 to use the interpreter, set the value of the variable
412 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
413 * minor incompatible change: the single-stepper REPL has been merged
414 with the normal debugger (see the "Stepping" heading of the debugger help
415 for more details). The debugger command STEP will no longer switch
416 to the single-stepper REPL.
417 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
418 for a type now works.
419 * bug fix: loading of default sysinit file works. (thanks to Leonid
421 * bug fix: better detection of circularities in the file-compiler.
422 (reported by Marco Monteiro)
423 * bug fix: the CL pathname functions now work with files that have
424 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
425 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
426 systems with Unix98 pty semantics.
427 * bug fix: ASDF-INSTALL will now work with bsd tar.
428 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
430 * bug fix: timers expiring in dead threads no longer cause a
431 type-error (reported by Paul "Nonny Mouse").
432 * bug fix: thanks to more lightweight single-stepper instrumentation,
433 code compiled with (DEBUG 3) will compile and execute significantly faster,
434 and will have more accurate type-inferencing than before
435 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
436 whose bindings are modified
437 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
438 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
439 ** replace dummy implementations of CL:MACHINE-INSTANCE and
440 CL:SOFTWARE-VERSION with proper version
442 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
443 * feature: implemented the READER-METHOD-CLASS and
444 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
445 as specified by AMOP.
446 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
448 * optimization: faster LOGCOUNT implementation on x86 and x86-64
449 (thanks to Lutz Euler)
450 * optimization: hashing of general arrays and vectors has been
451 improved. (reported by Any Fingerhut)
452 * enhancement: SB-INTROSPECT is now able to find definitions of
453 profiled functions. (thanks to Troels Henriksen)
454 * enhancement: compiler-macro expansion applies now to FUNCALL forms
455 as well, allowing compiler-macros for SETF-functions to expand.
456 * enhancement: step-instrumentation no longer wraps known
457 single-value functions with multiple-value context, allowing
458 better type inference.
459 * fixed bug #337: use of MAKE-METHOD in method combination now works
460 even in the presence of user-defined method classes. (reported by
461 Bruno Haible and Pascal Costanza)
462 * fixed bug #339(c): if there are applicable methods not part of any
463 long-form method-combination group, call INVALID-METHOD-ERROR.
464 (reported by Bruno Haible)
465 * fixed bug #361: the :FUNCTION initarg in the protocol for
466 initialization of methods can now be used to override
467 internally-produced optimized functions. (reported by Bruno
469 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
470 system-provided lambda expression no longer cause warnings about
471 unbound #:|pv-table| symbols.
472 * bug fix: improved the handling of type declarations and the
473 detection of violations for keyword arguments with non-constant
475 * bug fix: potentially erronous calls to PATHNAME and
476 MERGE-PATHNAMES were being flushed in some cases.
477 (reported by Richard Kreuter)
478 * bug fix: compiled calls to TYPEP were mishandling obsolete
479 instances. (reported by James Bielman and Attila Lendvai)
480 * bug fix: format strings with ~> without matching ~< no longer
481 trigger an AVER, but signal an understandable error instead.
482 (reported by Antonio Martinez)
483 * bug fix: specifying an output-file without a pathname-type for
484 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
485 of input-file instead of "fasl". (reported by Robert Dodier)
486 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
487 (reported by James Y Knight).
488 * bug fix: compiler-macros-function did not consider the environment
489 argument for shadowing by local functions.
490 * bug fix: compiler-macros expansion was inhibited by local INLINE
492 * bug fix: inline expansions of known functions were subject to
493 step-instrumentation in high debug policies, leading to problems
495 * bug fix: compiler failed to differentiate between different CONS
497 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
498 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
499 * bug fix: #368: incorrect use of expressed vs. upgraded array
501 * bug fix: #306a: more precise unions of array types.
502 * thread-safety improvements:
503 ** CONDITION-WAIT could return early on Linux, if the thread was
504 interrupted and subsequently continued with SIGCONT.
505 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
507 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
508 * added support for the ucs-2 external format. (contributed by Ivan
510 * minor incompatible change: pretty printing of objects of type
511 (cons symbol) is, in the default pprint-dispatch-table, now
512 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
514 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
515 called later in a class's lifetime, possibly as late as when the
516 first instance of the class is created. Previously,
517 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
518 class became finalizeable.
519 * fixed bug: FILE-POSITION sometimes returned inconsistent results
520 for multibyte external-format streams. (thanks to "vbzoli")
521 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
522 with :ALLOCATION :CLASS inherited from superclasses of the
524 * fixed bug: anonymous classes can now be created using the :NAME
525 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
526 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
527 * fixed bug: core-files saved with :EXECUTABLE T can again be
528 executed when SBCL_HOME isn't set. (reported by James Knight)
529 * fixed bug: toplevel LOCALLY forms with declarations could
530 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
531 * fixed bug: printing from several different threads using different
532 values of *print-case* could cause invalid output, due to
533 some internal special variables of the printer not being bound
534 thread-locally (reported by Max Mikhanosha)
535 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
536 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
537 (reported by Pascal Costanza)
538 * fixed bug: FINALIZE-INHERITANCE is called from
539 REINITIALIZE-INSTANCE on classes when the class has previously
540 been finalized, as required by AMOP.
541 * minor code generation optimizations:
542 ** better register allocation in CLOS dispatching functions
543 ** overflow detection when coercing signed bytes to fixnums on x86-64
544 ** is now implemented with one IMUL instruction instead of three shifts
545 ** more efficient bit-vector access on x86 and x86-64
546 ** more efficient access to raw structure slots on x86 and x86-64
547 * fixed some bugs revealed by Paul Dietz' test suite:
548 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
551 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
552 * feature: thread support on Solaris/x86, and experimental thread support
554 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
556 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
557 default method for SLOT-UNBOUND.
558 * minor incompatible change: prevent the user from specializing the
559 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
560 somewhat counter to the intent of the protocol, I (CSR) think, and
561 additionally it just doesn't work in SBCL as currently
562 implemented, thanks to optimizations (that are always valid for
563 the other three SLOT-VALUEish functions, but not for the setter).
564 * minor incompatible change: the :SB-LDB feature is now enabled by
565 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
566 the low-level debugger.
567 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
568 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
569 down to an SB-SYS:FD-STREAM, allowing direct communication in
570 more cases. Temporary files and pipes are still used as a fallback
572 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
573 on the non-reentrant C functions localtime(3) and gmtime(3).
574 * bug fix: global optimization policy was not visible in LOCALLY and
576 * bug fix: class objects can be used as specializers in methods.
577 (reported by Pascal Costanza)
578 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
579 with a type error. (reported by blitz_ on #lisp)
580 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
581 failed with a type error. (reported by Pascal Bourguignon)
582 * bug fix: merging pathnames against defaults with :DIRECTORY
583 starting with '(:RELATIVE :BACK) should preserve the :BACK.
584 (reported by James Y Knight)
585 * bug fix: saving large (>2GB) cores on x86-64 now works
586 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
587 constant index and a (simple-array (signed-byte 32)) array
588 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
589 error (patch by Robert J. Macomber)
590 * feature: TIME now displays the amount of run-time spent in GC
591 * bug fix: The debugger now does a better job of respecting
592 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
593 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
594 workaround for bug 403.)
595 * bug fix: floating point exception handling now works on FreeBSD
596 (thanks to NIIMI Satoshi)
597 * fixed some bugs revealed by Paul Dietz' test suite:
598 ** MISC.641: LET-conversion were not supposed to work in late
600 * improvements to the Win32/x86 port:
601 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
602 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
604 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
605 * new feature: source path information is generated for macro-expansion
606 errors for use in IDE's like Slime (thanks to Helmut Eller)
607 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
608 * bug fix: compiler does not loop forever on an invalid type in
610 * improvement: compilation of most CLOS applications is significantly
612 * optimization: added a limited bytecode compiler for simple toplevel
613 forms, speeding up compilation and FASL loading
614 * bug fix: the statistical profiler now properly distinguishes anonymous
617 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
618 * minor incompatible change: in sbcl-0.9.11 (but not earlier
619 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
620 into subclasses of the finalized class. Now user calls to
621 FINALIZE-INHERITANCE finalize just the one class, and calls by the
622 system return before any subclasses are finalized.
623 * minor incompatible change: The reader no longer ignores errors
624 regarding non-existent packages in #+ and #- feature tests.
625 * new feature: command line options --no-sysinit, --no-userinit to
626 inhibit loading the corresponding init files
627 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
628 for sending data through UDP sockets (thanks to François-René Rideau)
629 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
630 error is signalled instead (thanks to François-René Rideau)
631 * minor incompatible change: Error signalling behaviour of lexical
632 operations violating package locks has changed slightly. Refer to
633 documentation on package locks for details.
634 * enhancement: EVAL can process IF-expressions without resorting to the
636 * bug fix: Some operations involving SIN, COS, or TAN weren't being
637 constant-folded properly due to missing out-of-line functions.
638 (reported by Mika Pihlajamäki)
639 * bug fix: LISTEN sometimes returned T even in cases where no data was
640 immediately available from the stream
641 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
642 were reversed. (reported by Levente Mészáros)
643 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
644 when given a vector displaced to an adjustable array. (reported
646 * bug fix: derivation of float boundaries from numbers outside the
647 appropriate float range (reported by John Wiseman)
648 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
650 * bug fix: precomputing cache entries for generic functions with
651 some subclasses of specializers as yet invalid does not attempt to
652 fill a cache line with a negative offset. (reported by Levente
654 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
655 allow condition class objects as arguments to DOCUMENTATION and
656 (SETF DOCUMENTATION); only find and set documentation for
657 structure names for the STRUCTURE doc-type. (suggested by Gary
659 * improvements to the Win32/x86 port:
660 ** (user-homedir-pathname) and default initialization file
661 locations now know about the user's "Documents and Settings"
662 directory (thanks to Yaroslav Kavenchuk)
663 ** run-program is implemented (thanks to Mike Thomas)
664 ** sockets support (thanks to Timothy Ritchey)
665 ** better backtrace support (thanks to Alastair Bridgewater)
666 ** sb-grovel supported
667 ** asdf-install and sb-posix work somewhat
668 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
669 * improvements to the Solaris/x86 port:
670 ** works on Solaris 11/Solaris Express
671 ** floating-point exception handling support
672 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
673 * fixed some bugs revealed by Paul Dietz' test suite:
674 ** REMOVE-METHOD returns its generic function argument even when
675 the method is not one of the generic functions' methods.
676 ** objects declared as MEMBER types can be admissible arguments to
678 ** printing characters should simply be printed by the FORMAT ~:C
680 ** compiler failure when compiling functions with hairy constant
681 defaults for optional parameters.
682 ** compiler produces wrong code when MAYBE-INLINE-expanding a
683 function, which is already optimized.
685 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
686 * new platform: experimental support for SBCL x86/Darwin, including
687 MacOS X 10.4.5 on Intel.
688 * new feature: Unicode character names are now known to the system
689 (through CHAR-NAME and NAME-CHAR).
690 * new feature: the filesystem location of SBCL's core file is
691 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
692 * minor incompatible change: the contrib modules SB-POSIX and
693 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
694 this change is to make it easier to distribute
695 location-independent binaries.
696 * bug fix: as implied by AMOP, standardized classes no longer have
697 slots named by external symbols of public packages. (reported by
699 * optimization: calling functions via constant symbols -- as in
700 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
701 via the function object as in (FUNCALL #'FOO).
702 * optimization: on x86 and x86-64, the calling convention for the
703 case of dealing with an unknown number of values has been altered
704 to be friendlier to the prediction heuristics implemented,
705 particularly with reference to CALL and RET pairing. (thanks to
706 Alastair Bridgewater)
707 * optimization: on x86, the code for access to thread-local symbol
708 values and binding/unbinding of thread-local symbols is smaller.
709 (thanks to Alastair Bridgewater)
710 * enhancement: CONSTANTP is now able to determine constantness of
711 more complex forms, including calls to constant-foldable standardized
712 functions and some special forms beyond QUOTE.
713 * fixed bug: occasional GC crashes on Solaris/x86
714 * optimization: x86-64 supports stack allocation of results of simple
715 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
716 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
717 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
718 (thanks to James Knight)
719 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
720 by the standard (thanks to Stephen Compall)
722 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
723 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
724 be used for bundling the runtime and the core file into one
725 executable binary. This feature is not currently supported on all SBCL
726 platforms. (thanks to James Bielman and NIIMI Satoshi)
727 * new feature: a generational or ephemeral garbage collector is now
728 the default on the PowerPC platform (both Linux and Darwin). The
729 old Cheney (stop and copy) collector is a build-time option.
730 (thanks to Cyrus Harmon, after Raymond Toy)
731 * minor incompatible change: the method by which SBCL finds its
732 contributed modules has changed; it no longer relies on symbolic
733 links from an $SBCL_HOME/systems directory, but searches directly
734 in the subdirectories of $SBCL_HOME.
735 * enhancement: the dynamic heap size on the Linux/PPC platform is
736 markedly larger, even using the older Cheney garbage collector.
737 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
738 certain complicated string types. (reported by Gary King)
739 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
740 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
741 * fixed bug: DOCUMENTATION on structures no longer signals an
742 error if no structure type of the right name exists. (reported by
744 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
745 applicable non-standard primary or :AROUND method on
746 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
747 no longer cause unbound variable errors. (reported by Kevin Reid)
748 * fixed bug: metacircle resolution in cases where methods have slots
749 added before the slots from STANDARD-METHOD. (reported by Jean
751 * fixed bug: the Power PC assembly code for calling into lisp
752 sometimes computed its return address wrongly (depending on the
753 15th bit of the address, fixed at link time). This used to
754 manifest itself in a segmentation violation while building PCL.
755 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
756 many others over the years)
757 * fixed bug: ROOM no longer reports silly numbers for stack usage if
758 the stack pointer (treated as a signed integer) is negative.
759 (thanks to Peter van Eynde)
761 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
762 * new platform: experimental support for the Windows operating
763 system has been added. (thanks to Alastair Bridgewater)
764 * fixed several bugs in and robustified the PPC FFI (including
765 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
766 * bug fix: allow non-simple string symbol names (reported by Paul
768 * bug fix: interrupt handling on NetBSD (thanks to Richard M
770 * bug fix: saving a core corrupted callbacks on x86/x86-64
771 * bug fix: closed a loophole in metacircularity detection and
772 grounding in the PCL implementation of CLOS.
773 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
774 (reported by tomppa on #lisp)
775 * bug fix: clear the direction flag on Lisp -> C transitions, as
776 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
777 (reported by Andrej Grozin and Hendrik Maryns)
778 * optimization: major improvements to GC efficiency on GENCGC platforms
779 * optimization: faster implementation of EQUAL
780 * optimization: emit more efficient opcodes for some common
781 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
782 * optimization: several other minor code-generation improvements on x86-64
783 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
785 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
786 * minor incompatible change: (SETF CLASS-NAME) and (SETF
787 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
788 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
789 by ANSI as a generic function, and for consistency (SETF
790 GENERIC-FUNCTION-NAME) is treated likewise.
791 * fixed bug #233.b: make constraint propagation notice when a variable
792 value is changed after it is referenced but before it is used
793 * fixed bug #296: no more arbitrary behaviour from filenames with
794 odd characters as --load arguments. (reported by Adam Warner)
795 * fixed bug #390: :CHARACTER-SET pathname components now work as
796 expected. (reported by Tim Daly Jr)
797 * fixed bug #391: complicated :TYPE intersections in slot
798 definitions no longer cause an error in PCL internals.
799 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
800 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
801 returning the number of octets which would be written to the
802 file-stream. (thanks to Robert J. Macomber)
803 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
804 arguments to the internal stream functions. (thanks to David
806 * bug fix: stack exhaustion is now handled on the x86/SunOS
807 platform. (thanks to Daisuke Homma)
808 * bug fix: more accurate ROOM results on GENCGC platforms
809 * optimization: improved type inference for arithmetic-for
810 index variables in LOOP
811 * optimization: faster floating-point SQRT on x86-64
812 * fixed some bugs revealed by Paul Dietz' test suite:
813 ** DOCUMENTATION returns NIL instead of "" for method combinations
814 that don't have a docstring
816 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
817 * minor incompatible change: (SETF CLASS-NAME) and (SETF
818 GENERIC-FUNCTION-NAME) are no longer generic functions, and
819 therefore are not customizeable by user code (as seems to be at
820 least permitted and maybe required by AMOP). As a consolation,
821 however, the SBCL implementation of these functions now calls
822 REINITIALIZE-INSTANCE as specified by AMOP.
823 * bug fix: slot-definition documentation is propagated as per ANSI
824 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
825 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
826 argument, as do eslotd initialization methods. (from Pascal
827 Costanza's "Closer" project)
828 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
829 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
831 * bug fix: it is now possible to have more than one subclass of
832 STANDARD-GENERIC-FUNCTION without causing stack overflow.
833 (reported by Bruno Haible, Pascal Costanza and others)
834 * bug fix: the dependent update protocol now works for generic
835 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
837 * bug fix: condition-class instances corresponding to
838 DEFINE-CONDITION forms are now created eagerly. (reported by
839 Kalle Olavi Niemitalo on comp.lang.lisp)
840 * bug fix: floating point printing is more accurate in some
841 circumstances. (thanks to Simon Alexander)
842 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
843 merged with *DEFAULT-PATHNAME-DEFAULTS*.
844 * bug fix: callbacks on OS X now preserve stack-alignment.
845 (thanks to Cyrus Harmon)
846 * enhancement: the x86-64 disassembler is much better at
847 disassembling SSE instructions. (thanks to Lutz Euler)
848 * enhancement: CHECK-TYPE on a variable now allows type-inference
849 (thanks to Vincent Arkesteijn)
850 * optimization: improved performance of EUC-JP external format.
851 (thanks to NIIMI Satoshi)
852 * optimization: performance improvements to IO on file streams of
853 :ELEMENT-TYPE CHARACTER
854 * optimization: much faster memory allocation on x86-64
855 * optimization: faster garbage collections (latency and throughput)
857 * optimization: faster bignum subtraction and division on x86 and x86-64
858 (thanks to Lutz Euler)
859 * optimization: more accurate type-inference for loops that use a
860 floating point index variable or a negative step.
862 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
863 * bug fix: add a workaround to SBCL looping infinitely at startup on
864 Linux kernels with apparently buggy implementations of personality().
865 (thanks to Svein Ove Aas)
866 * bug fix: Unicode symbols are correctly printed in LDB backtraces
867 (thanks to David Lichteblau)
868 * bug fix: local bindings shadow global symbol macros for
869 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
870 * optimization: non-open coded uses of numeric comparison operators
871 (e.g. >) no longer cons when called with more than one parameter
872 on platforms supporting dynamic-extent allocation.
873 * enhancement: saving cores with foreign code loaded is now supported
874 on MIPS/Linux in addition to the previously supported platforms.
875 * bug fix: FIND-RESTART now tests for activity, not applicability when given
876 a restart object as identifier. (reported by Helmut Eller for CMUCL)
877 * bug fix: division by zero in sb-sprof when no samples were collected
878 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
879 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
880 (thanks to Alexander Kjeldaas)
881 * new feature: ignoring the return values of destructive functions
882 like NREVERSE signals a compile-time style-warning.
883 (thanks to Kevin Reid)
884 * new feature: the alignment of alien structure fields can be
885 explicitly specified. (thanks to Cyrus Harmon)
886 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
887 and dump core on SIGQUIT
889 ** incompatible change: threads do not inherit values of specials
890 from their parents (see manual)
891 ** bug fix: threads stacks belonging to dead threads are freed by the
892 next exiting thread, no need to gc to collect thread stacks anymore
893 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
894 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
895 ** bug fix: dynamic variable and thread start related gc lossage
896 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
898 * fixed some bugs revealed by Paul Dietz' test suite:
899 ** SUBTYPEP is slightly more accurate on heinously complicated
900 CONS types where some of the members have uncertain (in the
901 NIL, NIL sense) type relationships to each other.
903 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
904 no longer purified unless :PURIFY T is explicitly specified.
905 ** Non-purified cores are significantly smaller than before
907 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
908 * new feature: timers based on Zach Beane's excellent timer package
909 * added support for the following external formats: koi8-u,
910 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
911 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
912 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
913 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
914 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
915 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
916 * incompatible change: a threaded SBCL will no longer revert to
917 non-threaded mode on non-NPTL systems, but refuse to start entirely.
918 * bug fix: interrupts are disabled until startup is complete; no
919 more sigsegvs when receiving a signal to soon
920 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
922 * bug fix: add a workaround for the memory randomization features in
923 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
924 workaround will only be in effect on systems with the proc filesystem
926 * bug fix: printing objects of type HASH-TABLE signals a
927 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
929 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
930 have been running for over 50 days. (reported by Gilbert Baumann)
931 * bug fix: the logic for getting names of functions gets less
932 confused when confronded with alternate-metaclass
933 funcallable-instances. (reported by Cyrus Harmon)
934 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
935 consistent, even on internal alternate-metaclass objects.
936 * bug fix: flush closure information collected by physical
937 environment analysis prepass before the main pass. (bug reported
939 * bug fix: compiler pack phase does not modify a hashtable, which is
940 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
941 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
942 may be expanded. (reported by Alexander <malishev> on c.l.l)
943 * bug fix: no more style-warnings when compiling code using LAST
944 under high-SPEED low-SPACE settings. (reported by David Wallin
946 * fixed some incompatibilities between SBCL's MOP and the MOP
948 ** the METAOBJECT class is now implemented;
949 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
950 STANDARD-OBJECT, as required;
951 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
952 now compatible (as required by VALIDATE-SUPERCLASS); there
953 remains a consistency requirement over the presence or absence
954 of the FUNCTION class in the superclass list of finalized
955 classes; see the manual for more details;
956 ** the :AROUND method for COMPUTE-SLOTS on
957 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
958 requested slot ordering.
960 ** bug fix: parent thread now can be gc'ed even with a live
962 ** bug fix: binding a special with PROGV to no value is not
963 the same as that symbol not having been bound (thanks to
965 ** bug fix: binding specials is thread safe (thanks to
967 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
968 and SB-SPROF do not die with 'no handler for signal XX in
969 interrupt_handle_now(..)' anymore
970 ** bug fix: WITH-TIMEOUT works with multiple running threads
971 * fixed some bugs revealed by Paul Dietz' test suite:
972 ** ENSURE-GENERIC-FUNCTION should take a method class object for
973 the :method-class keyword argument.
975 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
976 * new port: the Solaris operating system on x86 processors is now
977 mostly supported, though some rough edges in the environment
978 remain. (thanks to Daisuke Homma)
979 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
980 space for its heap. (thanks to Thiemo Seufer)
981 * optimization: SBCL on MIPS platforms now supports dynamic-extent
982 closures. (thanks to Thiemo Seufer)
983 * minor incompatible change: eof selects abort in the debugger.
984 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
985 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
987 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
988 deprecated, and will go away in a future revision of SBCL.
989 * minor incompatible change: GC-ON and GC-OFF are no longer
990 implemented with a counter, it does not matter how many times gc
991 is switched on or off
992 * bug fix: discriminating functions for generic function classes
993 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
994 make invalid assumptions about method precedence order. (reported
996 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
997 platforms. (thanks to Thiemo Seufer)
998 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
999 * bug fix: correct pathname printing with printer escaping is on.
1000 (thanks to Kevin Reid)
1001 * bug fix: complex VOP definitions in "user-space" no longer trigger
1002 package locks. (reported by Zach Beane)
1003 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1004 for funcallable-instance functions, and is no different from
1006 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1007 characters in the current readtable. (reported by Nicholas Neuss)
1008 * bug fix: SUBTYPEP on various CONS types returns more a more
1009 accurate acknowledgment of its certainty.
1010 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1011 lists in some cases. This partially fixes bug 384.
1012 * flush all standard streams before prompting in the REPL and the
1014 * bug fix: signal handling and triggering gc do not conflict
1015 directly anymore, in particular a high frequency sb-sprof does
1016 not prevent gc from running
1017 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1018 approximation for timezone and DST information between the
1019 universal time epoch and the smallest negative 32-bit time_t.
1020 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1021 year 1899 so long as the encoded time is non-negative.
1022 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1023 the correct answer for non-constant rotate amounts when performing
1024 an inline 32-bit rotation.
1026 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1027 there is only one thread in the session
1028 ** bug fix: memory leak for streams created in one thread and
1029 written to in another
1030 ** bug fix: lockup when compiled with gcc4
1031 ** bug fix: race that allows the gc to be triggered when gc is
1033 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1034 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1036 ** bug fix: finished threads are gc'ed properly
1037 * fixed some bugs revealed by Paul Dietz' test suite:
1038 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1039 has arguments with a different set of applicable methods from
1040 the orignal arguments.
1041 ** The type error thrown by MAP now has a correct expected-type
1043 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1044 name a compiled function.
1045 ** Three MISC tests where a large bignum was improperly coerced to
1046 a float that couldn't represent that bignum during type
1047 derivation were fixed.
1048 ** SUBTYPEP can now handle types involving the negation of a
1049 list-form FUNCTION type.
1050 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1051 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1052 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1054 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1055 * New feature: Experimental support for bivalent streams: streams
1056 opened with :element-type :default now allow character and binary
1057 (unsigned-byte 8) I/O
1058 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1059 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1061 * bug fix: no more highly sporadic "couldn't check whether ~S is
1062 readable" when reading a stream and an interrupt hits in the middle
1063 of a select system call
1064 * compiler better recognizes complex arrays (reported by Eduardo
1066 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1067 for variables on SBCL built with linkage-tables. (reported by Luis
1069 * various error reporting improvements.
1070 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1071 (Thanks to Thiemo Seufer)
1072 * bug fix: nested reader invokations work correctly
1073 * bug fix: it is possible to have simultaneous references to foreign
1074 code and foreign data with the same name.
1076 ** added x86-64 support
1077 ** incompatible change: the threading api now works with thread
1078 objects instead of thread ids
1079 ** bug fix: threads are protected from signals and interruption when
1080 starting up or going down
1081 ** bug fix: a race where an exiting thread could lose its stack to gc
1082 ** bug fix: don't halt on infinite error in threads if possible
1083 ** fixed numerous gc deadlocks introduced in the pthread merge
1084 ** bug fix: fixed thread safety issues in read and print
1085 ** bug fix: debugger doesn't hang on session lock if interrupted at
1086 an inappropriate moment
1087 ** bug fix: run-program is now thread safe(r)
1088 ** bug fix: inner with-recursive-lock no longer releases the mutex
1089 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1090 * fixed some bugs revealed by Paul Dietz' test suite:
1091 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1092 EXPECTED-TYPE slots filled.
1093 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1095 ** MAKE-CONDITION accepts classes as type-designators.
1096 ** COMPILE may never return NIL.
1097 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1098 range before calling Unix time functions
1100 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1101 * numerous signal handling fixes to increase stability
1102 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1103 * minor incompatible change: we now correctly canonize default
1104 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1105 as per the MOP, rather than the historical (INITARG INITFUNCTION
1106 INITFORM). (reported by Bruno Haible)
1107 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1108 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1109 for more information.
1110 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1111 pathname is a directory pathname.
1112 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1113 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1115 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1116 :METACLASS argument in addition to a class name. (reported by
1117 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1118 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1119 (reported by Bruno Haible; patch from Gerd Moellmann)
1120 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1122 * bug fix: more cleanups to the floating point exception handling on
1123 x86-64 (thanks to James Knight)
1124 * bug fix: the compiler does not try to constant fold calls of
1125 COERCE and BIT-* functions when they return freshly constructed
1126 objects. (reported by Nikodemus Siivola and Paul Dietz)
1127 * optimization: improved the MIPS versions of generic (in the
1128 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1129 * optimization: direct conversion of (unsigned-byte 32) to floats on
1130 the PowerPC platform.
1131 * optimization: structure instances with raw slots now use less
1132 memory, and probably show better memory locality. (thanks to
1134 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1135 are treated as requests for stack allocation on the x86-64,
1136 Alpha-32, PPC and SPARC platforms.
1137 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1138 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1139 (thanks to Tony Martinez)
1141 ** gcing a dead thread can no longer lead to lockups
1142 ** threads block signals until they are set up properly
1143 ** errno is no longer shared by threads
1144 ** interrupt-thread restores the eflags register on x86
1145 ** fixed some lockups due to gc/thread interaction
1146 * fixed some bugs revealed by Paul Dietz' test suite:
1147 ** invalid dotted lists no longer raise a read error when
1148 *READ-SUPPRESS* is T
1149 ** signal an error if a symbol that names a declaration is used
1150 as the name of a type, or vice versa
1151 ** allow using the (declare (typespec var*)) abbreviation for
1152 (declare (type typespec var*)) with all type specifiers
1153 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1154 ** always return NIL from PPRINT-POP when OBJECT is NIL
1155 ** don't signal errors when pretty-printing malformed LABELS,
1156 FLET or MACROLET forms
1157 ** declarations in a DOLIST body are also in scope for the
1159 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1161 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1164 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1165 * fixed cross-compiler leakages that prevented building a 32-bit
1166 target with a 64-bit host compiler.
1167 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1168 opened with :IF-EXISTS :OVERWRITE.
1169 * fixed bug 281: error for an invalid qualifier in a short-form method
1170 combination method is not signalled until the faulty method is called.
1171 * bug fix: iteration variable type inferrer failed to deal with open
1172 intervals. (reported by Alan Shields)
1173 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1174 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1175 or more runtime options were provided to the sbcl binary.
1176 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1177 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1178 * optimization: there's now a fast-path for fixnum arguments in the
1179 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1180 * optimization: the code generated on x86-64 is more compact thanks
1181 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1182 * bug fix: floating-point exception handling now partly works on
1183 x86-64. (Thanks to James Knight)
1184 * improvement to the MIPS backend from Thiemo Seufer: C-style
1185 64-bit long long arguments and return values to alien functions
1187 * fixed some bugs revealed by Paul Dietz' test suite:
1188 ** the type-error signalled from WARN has a filled-in DATUM slot.
1189 ** the type-error required when a stream is not associated with
1190 a file has the stream as its datum.
1191 ** type-errors on single-floats on x86-64 no longer have
1192 :INVALID-OBJECT as the datum
1193 ** the type-errors signalled for invalid function names now have
1194 a correct expected type
1195 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1196 for structures defined with a :TYPE. Documentation strings for
1197 typed structures are no longer immediately discarded
1198 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1199 broadcast streams and synonym streams. FILE-LENGTH now also works
1200 on broadcast streams.
1202 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1203 * incompatible change: the --noprogrammer option, deprecated since
1204 version 0.7.5, has been removed. Please use the equivalent
1205 --disable-debugger option instead.
1206 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1208 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1209 inoperational for a while now) has been completely removed.
1210 * null lexical environments are now printed as #<NULL-LEXENV>,
1211 significantly reducing the amount of clutter in typical backtraces.
1212 * documentation on weak pointers, finalization, and after GC hooks
1213 has been added to the manual.
1214 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1215 as other specialized array types, is much faster. SUBSEQ and
1216 COPY-SEQ on such arrays have also been sped up.
1217 * optimization: EQL is now more efficient when at least other argument
1218 is known to be of type (OR FIXNUM (NOT NUMBER)).
1219 * fixed bug: the runtime is now less vulnerable to changes in the
1220 size of the SBCL object on OS X, and virtual memory is reserved for
1221 all spaces that need to be at a fixed address.
1222 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1223 * fixed bug: finalizers and after GC hooks that cause consing are now
1225 * fixed bug: compiler error messages and summaries are now printed to
1226 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1227 * fixed inference of the upper bound of an iteration variable.
1228 (reported by Rajat Datta).
1229 * fixed bug 211e: calling local functions with duplicated constant
1230 keyword argument no longer causes a bogus style warning about an
1232 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1233 ftype declaration to be disregarded. (reported by Dave Roberts)
1234 * fixed bug 373: caused by erronous compilation of references to alien
1235 variables in the runtime on ppc/darwin.
1236 * fixed bug 376: CONJUGATE type deriver.
1237 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1238 (reported by Baughn on #lisp)
1239 * a cross-compiler bug on non-x86 platforms has been identified and
1240 fixed. (thanks to Bruno Haible)
1241 * improvements to the MIPS runtime code for increased stability.
1242 (thanks to Thiemo Seufer)
1243 * increased the maximimum compact environment size to allow
1244 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1245 * partial workaround for bug 135: don't name the function we're
1246 calling for hairy cases of EVAL, so as not to accumulate one environment
1247 entry per call to EVAL. (reported by Kevin Reid)
1248 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1249 * optimization: type testing for non-vector arrays should be faster.
1250 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1252 * bug fix: setting 31st element of a bit vector to zero did not work
1254 * bug fix: redefining a class definition which failed due to a
1255 previous accessor / function clash now works (but see BUGS entry
1256 #380 for more problems in this area). (thanks to Zach Beane)
1257 * the long form of DEFINE-METHOD-COMBINATION disables method group
1258 checking when given a single method group with pattern *. (thanks
1260 * on x86 compiler supports stack allocation of results of simple
1261 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1262 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1263 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1264 * fixed some bugs related to Unicode integration:
1265 ** the restarts for recovering from input and output encoding
1266 errors only appear when there is in fact such an error to
1268 * fixed some bugs revealed by Paul Dietz' test suite:
1269 ** MISC.549 and similar: late transformation of unsafe type
1270 assertions into derived types caused unexpected code
1272 ** SCALE-FLOAT type deriver is less wrong.
1273 ** type derivers for EXP, LOG and similar functions compute result
1274 types for complex arguments better.
1275 ** (MISC.563) CONJUGATE type deriver works for very restricted
1277 ** out-of-line type testers for character strings are available.
1278 ** EQUAL compiler transform understands specialness of objects
1280 ** accessing double-floats stored on the stack now works on x86-64.
1281 ** debugger internals could sometimes create invalid lispobjs,
1282 resulting in GC crashes.
1283 ** MISC.548: type check weakening can convert required type into
1285 ** initialization forms for bindings are not in scope of free special
1288 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1289 * incompatible change: thread support for non-NPTL systems has
1290 been removed - locking is buggy and unreliable. A threaded
1291 SBCL build will now warn at startup and refuse to create
1292 new threads, unless futex support is detected in the kernel
1293 * incompatible change: the top level REPL now has only an ABORT
1294 restart associated with it, not TOPLEVEL and ABORT as it used to.
1295 TOP and TOPLEVEL are now available as debugger commands for
1296 returning to the top level.
1297 * incompatible change: forms evaluated in the REPL now use the
1298 global optimization policy.
1299 * incompatible change: user- and system-initialization files are
1300 no longer processed with LOAD, but by READ and EVAL; hence the
1301 global optimization policy, startup package, readtable, etc,
1303 * The .fasl file format number has been incremented because of
1304 various incompatible changes.
1305 * internal entry point details and argument counts no longer appear
1306 in backtraces unless explicitly requested by setting
1307 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1308 * built-in and standard functions no longer have names like "top
1309 level local call to FOO".
1310 * fixed bug 32: functions defined in non-null lexical environments
1311 now have more legible printed representation
1312 * fixed bug 33: functions defined in non-null lexical environemnts
1313 are now more amenable to inspection by INSPECT.
1314 * workaround for bug 354: XEPs no longer appear in backtraces unless
1315 explicitly requested.
1316 * fixed bug: receiving the signal which results from attempting to
1317 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1318 SIGBUS on other *BSDs) on architectures where the C stack is also
1319 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1320 to debugger confusion or wild instability. (reported by Cheuksan
1322 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1323 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1324 (reported by Lutz Euler)
1325 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1326 compiled code. (reported by Paul Dietz)
1327 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1328 specialized parameter to a method to something that is not TYPEP
1329 the specializer is now possible.
1330 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1331 face of package deletion.
1332 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1333 pathname. (reported by David Tolpin on c.l.l)
1334 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1335 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1336 STRING-STREAM and FILE-STREAM, respectively.
1337 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1338 than previously; in addition, its test suite is now run on build.
1339 (thanks to Luke Gorrie)
1340 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1341 * optimization: added a immediate representation for single-floats
1343 * fixed some bugs revealed by Paul Dietz' test suite:
1344 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1345 ** attempting to create a package with a colliding nickname causes
1346 correctable errors to be signalled.
1347 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1348 ** several x86-64 backend bugs related to sign-extension of immediate
1351 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1352 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1353 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1355 * fixed loading of multiply forward-referenced layouts.
1356 (thanks to Cheuksan Wang)
1357 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1358 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1359 (see COMPILE-FILE documentation for details of :PRINT option)
1360 * fixed bugs 19 and 317: fixed-format floating point printing is
1361 more accurate. This also fixes a bug reported by Adam Warner
1362 related to the ~@F format directive.
1363 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1365 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1366 dispatch table if the from-char is a dispatch macro character.
1367 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1368 by the wordsize no longer ignore the last word. (reported by Lutz
1370 * fixed bug in type checking of dynamic-extent variables. (reported
1372 * optimization: sequence traversal functions use their freedom to
1373 coerce function designators to functions.
1374 * optimization: code with many calls to CLOS methods specialized on
1375 CLOS classes has had redundant type checks removed.
1376 * optimization: added declarations to speed up operations that access
1377 the internal character database (for example STRING-UPCASE)
1378 * optimization: comparison operations between floats and sufficiently small
1379 fixnums no longer create extra rationals
1380 * fixed some bugs related to Unicode integration:
1381 ** portions of multibyte characters at the end of buffers for
1382 character-based file input are correctly transferred to the
1383 start of the buffer at the next read.
1384 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1385 passing it through to OPEN.
1386 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1387 argument given to internal calls to OPEN.
1388 ** the built-in comment readers (introduced by character sequences
1389 ";" and "#|") are more forgiving to encoding errors; they will
1390 STYLE-WARN and then attempt to resync the stream at a character
1391 boundary. (thanks to Teemu Kalvas)
1392 * fixed some bugs in the x86-64 port:
1393 ** Negative short int return values from c-calls are sign-extended
1395 ** The stack is aligned to 16-bytes for c-calls, as required by
1396 the ABI. (thanks to Cheuksan Wang)
1397 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1398 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1399 for enums. (thanks to Vincent Arkesteijn)
1400 ** Multiple small optimizations and bugfixes for floating point
1402 * fixed some bugs revealed by Paul Dietz' test suite:
1403 ** Space, Tab, Linefeed, Return and Page have the invalid
1404 secondary constituent character trait.
1405 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1407 ** WITH-INPUT-FROM-STRING should only update the index place on
1409 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1410 works more reliably.
1411 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1412 with one-dimensional arrays with fill pointers.
1413 ** TYPE-OF failed on a complex with an integer realpart and a
1415 ** compiler failure during type inference for the code of form
1416 (IF (EQL X (THE ...)) ...) (MISC.535).
1418 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1419 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1420 platform. The port supports 61-bit fixnums, large memory spaces
1421 and reloading shared object files.
1422 * enhancement: saving cores with foreign code loaded is now
1423 supported on ppc/Darwin in addition to the previously supported
1425 * enhancement: the statistical profiler now walks deeper into the
1426 call stack for more meaningful call-graphs and accrued time
1427 reports (x86/x86-64 only). It also now reports time spent in
1429 * enhancement: it is now possible to trace most individual methods
1430 of a generic function in addition to tracing the generic function
1432 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1434 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1435 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1436 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1437 produces an error. (thanks to Vincent Arkesteijn)
1438 * bug fix: NAMESTRING on pathnames with :WILD components in their
1439 directories works correctly. (thanks to Artem V. Andreev)
1440 * fixed bug 125: compiler preserves identity of closures. (reported
1442 * bug fixed: functions with &REST arguments sometimes failed with
1443 "Undefined function" when compiled with (DEBUG 3). (reported by
1445 * bug fix: overflow during compiling of setting element of a bit
1446 vector with constant index and value. (reported by Timmy Douglas)
1447 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1448 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1449 * on x86 compiler partially supports stack allocation of dynamic-extent
1451 * GO and RETURN-FROM do not check the extent of their exit points
1452 when compiled with SAFETY 0.
1453 * fixed some bugs related to Unicode integration:
1454 ** encoding and decoding errors are now much more robustly
1455 handled; it should now be possible to recover even from invalid
1456 input or output to the terminal. (thanks to Teemu Kalvas)
1457 ** provided a first cut at implementing STRING-TO-OCTETS and
1458 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1459 ** altered the SB-MD5 contributed module slightly, changing the
1460 interface just enough for it to be supportable for builds where
1461 lisp characters are not eight bits.
1462 * fixed some bugs revealed by Paul Dietz' test suite:
1463 ** the FORMATTER-generated functions for ~V[ conditionals require
1464 the correct number of arguments.
1465 ** READ-FROM-STRING returns the mandated second value when applied
1466 to displaced strings.
1467 ** the #\Rubout and #\Backspace characters are treated as invalid
1468 constituent characters by the tokenizer.
1470 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1471 * new feature: reloading changed shared object files with
1472 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1473 * new feature: references to foreign variables and functions
1474 can now be compiled and loaded before the corresponding shared
1475 object file is loaded, as long as the foreign definitions are
1476 available at runtime.
1477 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1478 supported. (thanks to Dan Debertin)
1479 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1480 just keywords. (thanks to Vincent Arkesteijn)
1481 * fixed bug #331: structure-class instances corresponding to
1482 DEFSTRUCT forms are now created eagerly.
1483 * fixed bug #345: backtraces from calls to undefined functions work
1484 on x86 as well. Related bug #61 is now also partially fixed on x86
1485 (backtraces from throws to unknown catch tags.)
1486 * bug fix: lambda-list parsing is now stricter vrt. order and number
1487 of lambda-list keywords.
1488 * bug fix: as specified by AMOP, an error is signalled if a
1489 class-option appears multiple times in a DEFCLASS form. (reported
1491 * bug fix: RANDOM can be compiled when the compiler derives the type
1492 of its numeric argument as a disjoint set of small integers.
1493 (reported by Paul Dietz)
1494 * bug fix: starting a core saved with shared objects loaded when
1495 those objects are not available no longer causes threaded SBCL to
1496 hang. (reported by Sean Ross)
1497 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1499 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1500 argument, which must always be NIL. (reported by Kalle Niemitalo)
1501 * bug fix: printing 1.0d+23 no longer results in an error.
1502 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1503 * bug fix: structure slot setters preserve evaluation order. (thanks
1505 * fixed some bugs related to Unicode integration:
1506 ** RUN-PROGRAM can allow its child to take input from a Lisp
1507 stream. (reported by Stefan Scholl)
1508 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1509 locales. (reported by Ken Causey)
1510 * fixed some bugs revealed by Paul Dietz' test suite:
1511 ** INCF, DECF and REMF evaluate their place form as specified in
1513 ** FORMATTER expands ~{ ~} iteration directives with V or #
1514 parameters correctly.
1515 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1516 consequent uses no arguments correctly.
1517 ** the system has a partial understanding of the (COMPLEX RATIO)
1519 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1522 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1523 * new feature: a build-time option (controlled by the :SB-UNICODE
1524 keyword feature, enabled by default) for building the system with
1525 support for the entire 21-bit character space defined by the
1527 * new feature: the system now has rudimentary external-format
1528 support; the primary user-visible change at this time is that
1529 characters with the high bit set (such as lower-case-e-acute) will
1530 print correctly to a terminal in a UTF-8 environment.
1531 * minor incompatible change: BASE-CHAR no longer names a class;
1532 however, CHARACTER continues to do so, as required by ANSI.
1533 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1534 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1535 has been moved to the SB-EXT package (temporarily re-exported from
1537 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1538 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1540 * fixed bug #351: better error-handlind and reporting for malformed
1542 * fixed bug #350: bignum-printing is now more memory-efficient,
1543 allowing printing of very large bignums, eg. (expt 2 10000000).
1544 (reported by Bruno Haible)
1545 * fixed bug #302: better primitive-type selection for intersection
1547 * fixed bug #308: non-graphic characters now all have names, as
1548 required. (reported by Bruno Haible)
1549 * bug fix: redefining a class with different superclasses now correctly
1550 removes it from the direct-subclasses of its previous superclasses.
1551 (reported by David Morse)
1552 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1553 new value now works. (reported by Bruno Haible)
1554 * bug fix: correct canonicalization of multiple non-standard slot
1555 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1556 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1557 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1558 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1559 now exists, an signals an error.
1560 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1561 during macroexpansion for non-iteration variables as well. (reported
1562 by Bruno Haible for CMUCL)
1563 * bug fix: Cyclic structures and unprintable objects in compiler
1564 messages no longer cause errors. (reported by Bruno Haible)
1565 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1566 and READ-FROM-STRING all now return a primary value of NIL if
1567 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1568 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1569 (reported by Bruno Haible for CMUCL)
1570 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1571 do not designate a condition. (reported by Bruno Haible for
1573 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1574 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1575 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1576 permitting resolution in favour of any of the conflicting symbols.
1577 (reported by Bruno Haible for CMUCL)
1578 * bug fix: EQUAL compiler optimizations is less aggressive on
1579 strings which can potentially compare true despite having distinct
1580 specialized array element types.
1581 * bug fix: unit enumerations can be defined without dividing by
1582 zero. (thanks to Vincent Arkesteijn)
1583 * FORMAT compile-time argument count checking has been enhanced.
1584 (report from Bruno Haible for CMUCL)
1585 * a partial workaround for the bug 262: the compiler does not try to
1586 inline-expand a local function doing RETURN-FROM from a deleted
1587 BLOCK. (thanks to Peter Denno for the bug report and to David
1588 Wragg for the simple test case)
1589 * fixed some bugs revealed by Paul Dietz' test suite:
1590 ** RENAME-PACKAGE allows all package designators as new package
1592 ** constraint propagation and lambda variable substitution are
1593 more cautious in dealing with partially deleted code.
1594 ** compiler sometimes lost reoptimization passes.
1595 ** CERROR, given a condition as condition designator, passes its
1596 remaining arguments to the continue format control without
1598 ** Case-altering FORMAT directives work correctly on non-ASCII
1600 ** The REAL type specifier handles bounds outside the floating
1601 point ranges without signalling FLOATING-POINT-OVERFLOW.
1602 ** Functions with IR1-transformations can create intercomponent
1603 references to global functions.
1604 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1606 ** FORMAT ~R treats a nil value for its first parameter correctly.
1608 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1609 * enhancement: saving cores with foreign code loaded is now
1610 supported on x86/NetBSD and sparc/Linux in addition to the previously
1611 supported platforms.
1612 * bug fix: on some platforms repeated installations caused multiple
1613 copies of HTML documentation to be installed -- should not happen
1614 any more. (reported by Stefan Scholl)
1615 * bug fix: parsing self-recursive alien record types multiple times
1616 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1617 original patch by Helmut Eller for CMUCL)
1618 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1619 (thanks to Richard Kreuter)
1620 * bug fix: defining classes whose accessors are methods on existing
1621 generic functions in other (locked) packages no longer signals
1622 bogus package lock violations. (reported by François-René Rideau)
1623 * bug fix: special variables as DEFMETHOD parameters no longer have
1624 associated bogus type declarations. (reported by David Wragg and
1626 * bug fix: read-write consistency on streams of element-type
1627 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1628 * bug fix: redefiniton of the only method of a generic function with
1629 no DEFGENERIC no longer emits a full WARNING. In addition,
1630 redefinition of generic functions with no DEFGENERIC to an
1631 incompatible lambda list now signals an error. (thanks to Zach
1633 * bug fix: DEFGENERIC now works even when there's a function of the
1634 same name in an enclosing lexical environment. (thanks to Zach
1636 * fixed compiler failure, caused by instrumenting code during
1637 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1638 * optimization: added loop analysis and improved register allocation
1639 * fixed some bugs revealed by Paul Dietz' test suite:
1640 ** POSITION on displaced vectors with non-zero displacement
1641 returns the right answer.
1642 ** (SIMPLE-STRING) is a valid type specifier for sequence
1644 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1646 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1647 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1649 ** The justification version of the FORMAT ~< directive treats
1650 non-zero minpad parameter correctly.
1652 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1653 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1654 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1655 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1656 the supported interface.
1657 * new feature: Single-stepping of code compiled with DEBUG 2 or
1658 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1659 * new feature: saving cores with foreign code loaded is now
1660 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1661 Timothy Moore's work for CMUCL)
1662 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1663 parameters to * instead of NIL if no initform is supplied.
1664 (reported by Johan Bockgaard)
1665 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1666 correctly when FUNCALL appears as the car of the form. Note:
1667 despite this FUNCALL forms are not currently subject to
1668 compiler-macro expansion. (port of Raymond Toy's fix for the
1669 same from CMUCL, reported by Johan Bockgaard)
1670 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1671 (thanks for Teemu Kalvas)
1672 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1673 case instead of terminating the iteration (thanks for Julian
1674 Squires, Sean Champ and Raymond Toy)
1675 * bug fix: incorrect expansion of defgeneric that caused a style
1676 warning. (thanks for Zach Beane)
1677 * bug fix: slot accessor effective method computation works properly
1678 for classes with multiple non-standard applicable methods on
1679 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1680 * on x86 compiler supports stack allocation of results of LIST and
1681 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1682 CMUCL implementation by Gerd Moellmann)
1683 * fixed some bugs revealed by Paul Dietz' test suite:
1684 ** FORMAT strings with both the ~<~:;~> form of the justification
1685 directive and pretty-printing directives cause an error.
1687 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1688 * incompatible change: the internal functions
1689 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1690 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1691 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1692 instead of the old functions.
1693 * new feature: on platforms where "dladdr" is available foreign
1694 function names now appear in backtraces. (based on Helmut Eller's
1696 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1697 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1699 * documentation: ASDF manual is now distributed with SBCL.
1700 * bug fix: SBCL can now load its contributed modules with REQUIRE
1701 even if the system-provided entries have been removed from
1702 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1704 * bug fix: backtraces involving undefined functions or assembly
1705 routines are more informative. (thanks to Brian Downing)
1706 * bug fix: mutually referent alien structures now work correctly.
1707 (reported by Rick Taube)
1708 * bug fix: structures defined by WITH-ALIEN can be referred to
1709 within other definitions in the same WITH-ALIEN.
1710 * bug fix: division operators (MOD, TRUNCATE and the like) with
1711 constant zero divisors and integer dividends no longer generate
1713 * bug fix: provide default methods for INPUT-STREAM-P and
1714 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1715 * bug fix: improve the detection and resolution of MOP metacycles
1716 (where effective method computation depends on the generic function
1717 having its effective method computed). (reported by Bruno Haible)
1718 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1719 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1720 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1721 * bug fix: pathnames with relative directory components can be
1722 represented relative to default pathnames.
1723 * optimization: in taking the GCD of bignums, reduce the two bignums
1724 to approximately the same size (using Euclid's algorithm) before
1725 applying the more sophisticated binary GCD. (thanks to Juho
1727 * optimization: COUNT on bitvectors now operates word-at-a-time.
1728 * optimization: ASH with a positive, but not necessarily constant,
1729 (leftwards) shift, when in a modular context, is compiled to a
1731 * fixed some bugs revealed by Paul Dietz' test suite:
1732 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1733 if the corresponding argument is NIL.
1734 ** FORMAT directives accepting a minpad parameter treat negative
1736 ** literal commas as character parameters to FORMAT directives are
1738 ** literal spaces directly after ~<Newline> directives within a
1739 format-logical-block (~:< ~@:>) do not induce :FILL-style
1740 conditional newlines.
1741 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1742 as specified: it no longer includes conditional newlines).
1743 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1745 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1746 :PER-LINE-PREFIX argument does not evaluate to a string.
1748 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1749 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1750 the manual for details; for now, package locks can be disabled by
1751 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1752 if no major problems are found then it is likely that they will be
1753 compiled in unconditionally.
1754 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1755 unsupported operators on all platforms. To load a shared library
1756 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1757 object file, link it into a shared library outside of SBCL and
1758 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1760 * bug workaround: ROOM T and the GC were not working together
1761 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1762 reporting facility aren't preserved. That reporting has been
1763 disabled, so now until and unless someone figures out how to make it
1764 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1765 * minor incompatible change: as threatened around sbcl-0.8.0, the
1766 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1767 The symbols are also exported from SB-PCL for backwards
1768 compatibility, but more so than before SB-PCL should be treated as
1769 an implementation-internal package.
1770 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1772 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1773 It is known as of this release not to work on the Alpha, however.
1774 * fixed bug #167: errors signalled due to illegal syntax in method
1775 bodies are now more legible.
1776 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1777 designators and can hence be used with TYPEP.
1778 * fixed bug #333: CHECK-TYPE now ensures that the type error
1779 signalled, if any, has the right object to be accessed by
1780 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1781 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1782 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1784 * fixed bug #334: programmatic addition of slots using specialized
1785 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1786 and :ALLOCATION :CLASS slots.
1787 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1788 float exponents. (rereported by Peter Seibel)
1789 * fixed a bug: #\Space (and other whitespace characters) are no
1790 longer considered to be macro characters in standard syntax by
1791 GET-MACRO-CHARACTER.
1792 * fixed bug: initialization of condition class metaobjects no longer
1793 causes an instance of the condition to be created. (reported by
1795 * fixed bug: it is now possible to have slots such that
1796 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1797 description is neither :INSTANCE nor :CLASS.
1798 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1799 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1800 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1801 non-standard SLOT-DEFINITION-ALLOCATION values.
1802 * fixed bugs: various race conditions handling exiting threads.
1803 CL-PPCRE's thread torture test now passes.
1804 * fixed arguably-a-bug: GC time proportional to number of threads in
1805 system even when most of them are idle
1806 * optimization: improved performance of BIT and SBIT on bit-vectors.
1807 * .fasl file incompatibility: The fasl file version number has
1808 been incremented because of changes associated with package locks.
1810 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1811 * minor incompatible change: the system no longer provides
1812 optimization hints (conditions which are TYPEP
1813 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1814 mode; these hints are still emitted when higher SPEED optimization
1816 * new contrib module: a sampling profiler (profiling by statistical
1817 sampling, rather than by instrumenting functions) is available as
1818 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1819 * the behaviour of the standard function ED is now customizeable by
1820 third parties through a hook variable: see ED's documentation
1821 string for information on the protocol.
1822 * the compiler no longer emits efficiency notes for (FUNCALL X)
1823 when the type of X is uncertain under default optimization
1825 * fixed bug 276: mutating a binding of a specialized parameter to a
1826 method to something that is not TYPEP the specializer is now
1828 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1829 exhibit the required behaviour on the x86 platform. (thanks to
1830 Peter van Eynde, Eric Marsden and Bruno Haible)
1831 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1832 even for difficult arguments. (reported by Peter Graves)
1833 * fixed bug 141a: the backquote printer now descends quoted
1835 * fixed another bug in backquote printing: no more destructive
1836 modification of the form's list structure. (reported by Brian
1838 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1839 move between its address being taken and the call to
1840 interrupt_thread, fixing a crashing race condition.
1841 * the SB-POSIX contrib implementation has been adjusted so that it
1842 no longer exhibits ridiculously poor performance when constructing
1843 instances corresponding to C structs.
1845 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1846 * minor incompatible change: the sb-grovel contrib now treats C
1847 structures as alien (in the sense of SB-ALIEN) objects rather than
1848 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1849 has implications for memory management of client code
1850 (sb-grovel-returned objects must now be manually managed) and for
1851 type safety (alien objects now have full types).
1852 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1853 used to control emission of compiler diagnostics, rather than the
1854 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1855 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1856 quality should be considered deprecated.
1857 * install.sh now installs the user manual as well
1858 * (not quite a new documentable feature, but worth considering in
1859 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1860 beginnings of a semantically meaningful condition hierarchy is
1861 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1863 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1864 designator as the defaults argument.
1865 * fixed bug: Displaced arrays whose displaced-to array has become
1866 too small now cause ARRAY-DIMENSION to signal an error, providing
1867 for safer bounds-checking. (reported by Bruno Haible)
1868 * fixed bug: DEFCLASS slot definitions with identical :READER and
1869 :WRITER names now signal a reasonable error. (reported by Thomas
1871 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1873 * fixed bug: Invalid slot specification errors now print correctly.
1874 (thanks to Zach Beane)
1875 * fixed bug 320: Shared to local slot value transfers in class
1876 redefinitions now happen corrently from superclasses as
1877 well. (reported by Bruno Haible)
1878 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1880 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1881 improper lists correctly. (reported by Bruno Haible)
1882 * fixed bug 313: source-transform for <fun-name> was erroneously
1883 applied to a call of a value of a variable with name <fun-name>.
1884 (reported by Antonio Menezes Leitao)
1885 * fixed bug 307: The obsolete instance protocol ensures that
1886 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1887 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1888 remove dead unknown-values globs from the middle of the stack.
1889 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1890 easier to resume long computations after using *BREAK-ON-SIGNALS*
1891 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1892 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1893 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1894 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1895 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1896 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1898 * fixed a (fairly theoretical) bug in string printing: if
1899 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1900 does not have array-element-type equal to the most general string
1902 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1903 function name. (thanks to Nikodemus Siivola)
1904 * fixed bug: compile-time format string checker failed on
1905 non-closed ~{. (reported by Thomas F Burdick)
1906 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1907 don't warn on legal loop code involving more than one aggregate
1908 boolean. (thanks to Nikodemus Siivola)
1909 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1910 in keyword position to MAKE-PACKAGE was not being reported as
1912 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1913 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1915 * fixed bugs 280 and 312: the checking for multiple definitions in a
1916 file is less likely to become confused by uses of inline
1918 * fixed bug: the #S reader macro performs the keyword coercion
1919 specified for slot names. (reported by Kalle Niemitalo)
1920 * fixed bug: lambda lists may contain symbols whose names start with
1921 & but are not lambda-list-keywords; their occurrence triggers a
1923 * fixed bug 321: define-method-combination argument lambda lists do
1924 not cause invalid code to be generated when &optional or &aux
1925 variables have default values. (reported by Bruno Haible)
1926 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1927 have CLOS classes; this appears also to have fixed reported
1928 instability in stack exhaustion detection.
1929 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1930 MUFFLE-WARNING when no associated restart is present is now
1932 * optimization: rearranged the expansion of various defining macros
1933 so that each expands into only one top-level form in a
1934 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1936 * optimization: used a previously little-used slot in symbols to
1937 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1939 * fixed some bugs revealed by Paul Dietz' test suite:
1940 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1942 ** Fixed an optimization bug related to inheritance of initforms
1943 from local to shared slots.
1944 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1945 ** CAST optimizer forgot to flush argument derived type.
1946 ** print/read consistency on floats is now orders of magnitude
1947 more likely. (thanks also to Bruno Haible for a similar report
1949 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1950 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1951 for objects of type REAL. Make it so.
1952 ** SXHASH is specified (infelicitously) to respect similarity,
1953 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1954 it so. (thanks to Markus Ziegler)
1955 ** on the Alpha, the compiler succeeds in compiling functions
1956 returning a known number of arguments greater than 63.
1957 ** fixed handling of invalid NIL arguments in keyword position
1959 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1961 ** provided a readably-printable representation for RANDOM-STATE
1963 ** ensured that pathnames, where they have a namestring, always
1964 print using #P"..." syntax.
1966 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1967 * Support for the forthcoming 2.0 version of the NetBSD kernel
1968 running on x86 hardware has been added. (thanks to Perry
1969 E. Metzger most immediately, and others for their past work)
1970 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1971 current release version; SBCL's OpenBSD support had been broken
1972 since about the time of OpenBSD's switch to ELF binary format.)
1973 * [placeholder for DX summary]
1974 ** user code with &REST lists declared dynamic-extent, under high
1975 speed or space and low safety and debug optimization policy.
1976 * The manual has been converted to Texinfo format and the debugger
1977 chapter from the cmucl manual has been added.
1978 * A facility has been added to extract documentation strings from
1979 sbcl and store them as Texinfo-formatted snippets for inclusion in
1980 the manual (via Texinfo's @include directive)
1981 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1982 the test case to Dave Roberts)
1983 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1984 pointers. (reported by Sean Ross)
1985 * bug fix: PROFILE output is printed nicely even for large numerical
1986 values. (thanks to Zach Beane)
1987 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1988 greater than 32 handle EOF correctly.
1989 * bug fix: on X86 an immediate argument of the IMUL instruction is
1990 correctly printed in disassembly. (thanks to Lutz Euler)
1991 * bug fix: class slots in redefined classes preserve their old
1992 values. (thanks to Bruno Haible and Nikodemus Siivola)
1993 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1994 succeeds. (reported by Marco Baringer)
1995 * fixed some bugs revealed by Paul Dietz' test suite:
1996 ** READ-SEQUENCE now works on ECHO-STREAMs.
1997 ** RATIONALIZE works more according to its specification. (thanks
2000 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2001 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2002 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2003 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2004 should matter to you only if you rebind the printer control
2005 variables and then find you want different bindings in the
2006 debugger than in the ordinary execution of your program.)
2007 * The runtime build system has been tweaked to support building
2008 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2009 assembler and linker. (thanks to Nikodemus Siivola)
2010 * Unbound, undefined, undeclared variables now trigger full
2011 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2012 more often programmer error than deliberate exploitation of undefined
2014 * optimization: the hash algorithm for strings has changed to one
2015 that is less vulnerable to spurious collisions. (thanks to Juho
2017 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2018 needless bounds checking. (thanks to Juho Snellman)
2019 * optimization: implemented multiplication as a modular
2020 (UNSIGNED-BYTE 32) operation on the PPC backend.
2021 * fixed some bugs revealed by Paul Dietz' test suite:
2022 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2024 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2025 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2026 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2028 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2030 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2032 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2033 their output stream on EOF from read.
2034 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2035 have been read to end-of-file.
2036 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2038 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2039 description of determination of which consecutive characters
2041 ** Printing the "Space" character with escaping on now yields "#\\ ",
2042 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2043 ** Reading floating-point numbers with *READ-BASE* set to a number
2044 less than 10 works correctly.
2045 ** Reading floating-point numbers with *READ-BASE* set to a number
2046 more than 10 works correctly.
2047 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2048 the readtable currently in effect.
2050 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2051 * minor incompatible change: parsing of namestrings on a physical
2052 (Unix) host has changed; numbers after the final #\. in a
2053 namestring are no longer interpreted as a version field. This is
2054 intented to be largely invisible to the user, except that the
2055 meaning of the namestring "*.*.*" has changed: it now refers to a
2056 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2057 should usually be replaced by
2058 (make-pathname :name :wild :type :wild :version :wild)
2059 with the added benefit that this is more likely to be portable.
2060 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2061 option to OPEN now signals an error if the file being opened
2062 exists; this may have an impact on existing code.
2063 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2064 signals received, on the PowerPC platforms (both Linux and
2065 Darwin). (thanks to Pierre Mai for pointing out the location of
2067 * several fixes on OS X: The system now builds and runs cleanly on
2068 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2069 (thanks to Brian Mastenbrook)
2070 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2071 second-resolution: integer multiples of 1/3600 between -24 and 24.
2072 (thanks to Vincent Arkesteijn)
2073 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2074 type when called with 1 argument; PEEK-CHAR checked type of
2075 PEEK-TYPE only after having read first character from a
2076 stream. (reported by Peter Graves)
2077 * bug fix: the garbage collector now has much better locality
2078 behaviour, and in particular no longer treats all memory as being
2079 exhausted when many small objects point to each other in a deeply
2081 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2082 recognized as being TYPEP their class.
2083 * bug fix: the PUSHNEW documentation string has been corrected.
2084 (thanks to Vincent Arkesteijn)
2085 * bug fix: defaulting of the value for the last of an atypically
2086 large number of multiple values being bound was not being
2087 performed correctly on the Alpha or PPC platforms
2088 * optimization: implemented multiplication as a modular
2089 (UNSIGNED-BYTE 32) operation on the x86 backend.
2090 * optimization: SEARCH on simple-base-strings can now be open-coded.
2091 (see also contrib/compiler-extras.lisp for inspiration for
2092 teaching the compiler about the Boyer-Moore algorithm).
2093 * value, returned by MAX (and MIN) called with several EQUALP, but
2094 not EQL, arguments now does not depend on compiler settings.
2095 * fixed some bugs revealed by Paul Dietz' test suite:
2096 ** in stack analysis liveness information is propagated from
2097 non-local entry points.
2098 ** pathwise CAST removing failed when the CAST node did not start
2100 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2101 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2103 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2104 host is already defined.
2105 ** RENAME-FILE works on streams instead of signalling an internal
2107 ** PEEK-CHAR uses the current readtable when determining whether
2108 or not a character is whitespace.
2109 ** MERGE-PATHNAMES handles the case when the pathname does not
2110 specify a name while the default-pathname specifies a version.
2111 ** Pathnames now stand a better chance of respecting print/read
2113 ** Attempting to use standardized file system operators with a
2114 pathname with invalid :DIRECTORY components signals a
2116 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2118 ** DIRECTORY on logical pathnames is more correct.
2119 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2120 signal a TYPE-ERROR if their argument is not a stream.
2121 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2122 designator argument does not designate a stream.
2123 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2124 examining the synonym.
2125 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2127 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2128 element-type larger than ([UN]SIGNED-BYTE 32).
2130 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2131 * When built with the :SB-FUTEX feature, threaded builds now take
2132 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2133 for faster/more reliable mutex and condition variable support.
2134 * Incompatible change (but one you probably shouldn't have been using
2135 anyway): the interface and code for arbitrating between multiple
2136 threads in the same user session has been redesigned.
2137 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2138 error when called without an explicit environment argument.
2139 (thanks to Nikodemus Siivola)
2140 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2141 confused about their position. (thanks to Adam Warner and Gerd
2143 * bug fix: Pretty printing backquoted forms with unquotations in the
2144 argument list position of various code constructs such as LAMBDA
2145 now works correctly. (reported by Paul Dietz)
2146 * bug fix: Pretty printing unquotations no longer loses all
2147 stream position information.
2148 * optimization: performance of string output streams is now less
2149 poor for multiple small sequence writes.
2150 * optimization: performance of CSUBTYPEP in the presence of complex
2151 expressions involving CONS and NOT many times has been improved.
2152 (reported by Paul Dietz)
2153 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2155 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2156 streams for the REPL, for the debugger, and so on. (thanks to
2158 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2159 * fixed some bugs revealed by Paul Dietz' test suite:
2160 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2161 is ignored if it is too small, rather than propagating through
2162 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2163 ** extremely complex negations of CONS types were not being
2164 sufficiently canonicalized, leading to inconsistencies in
2166 ** VALUES tranformer lost derived type.
2168 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2169 * fixed a bootstrapping bug: the build process no longer assumes
2170 that the various BOOLE-related constants have the same value in
2171 host and target lisps. (noted by Paul Dietz' test suite on an
2172 SBCL binary built from CLISP)
2173 * The system can now be dynamically linked on the MIPS platform,
2174 which enables dynamic loading of foreign code from Lisp. (thanks
2175 to Ralf Baechle for discussions on the MIPS ABI)
2176 * The system now records debugging information for its own source
2177 files in a filesystem-position-independent manner, relative to
2178 the "SYS" logical host.
2179 * fixed a compiler bug: MV-LET convertion did not check references
2180 to the "max args" entry point. (reported by Brian Downing)
2181 * tweaked disassembly notes to be less confident about proclaiming
2182 some instruction as an LRA. (thanks to Brian Downing)
2183 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2184 now each have their own history, command character, and other
2185 characteristics. (thanks to David Lichteblau)
2186 * fixed some bugs revealed by Paul Dietz' test suite:
2187 ** compiler failure in compiling LOGAND expressions including a
2189 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2190 shift greater than 32.
2191 ** FLUSH-DEST did not mark blocks for type check regeneration.
2192 ** HANDLER-CASE failed to accept declarations in handler clauses
2193 in some circumstances.
2195 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2196 * New code in contrib/sb-introspect (still probably not entirely
2197 stable yet) provides some support for smart Lisp development
2198 environments like SLIME.
2199 * The conditions signalled for errors occurring when loading .fasl
2200 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2201 in a way which should help ASDF recover gracefully.
2202 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2203 stuff has been cleaned up. If you code contrib/ stuff, this might
2204 affect you, and you can look at contrib/README, contrib/STANDARDS,
2205 and/or the 0.8.4.27 diff to check.
2206 * In full calls the compiler now does not generate checks for declared
2207 argument types for all arguments.
2208 * various threading fixes
2209 ** and some experimental patches which didn't make it into
2210 the main tree for this release, but which are shipped in
2211 contrib/experimental-thread.patch as a possible fix for some
2212 failures (deadlock, spinning...) in GC-intensive multithreaded
2214 * fixed PPC build problem (source code incompatibility of different
2215 library versions): added offsetof() hackery which attempts to divine
2216 where glibc maintainers put uc_mcontext today
2217 * fixed bug 282: compiler does not trust type assertions while passing
2218 arguments to a full call.
2219 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2220 &OPTIONAL VALUES type parameter.
2221 * fix bug 214: algorithm for noting rejected templates is now more
2222 similar to that of template seletion. (also reported by rydis on
2224 * fixed bug 141b: printing backquoted information readably and prettily
2225 inserts a space where necessary.
2226 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2227 (CHARACTER 10) are now reported as errors, rather than propagated
2228 as unknown types. (reported by piso on #lisp)
2229 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2230 with values NIL and :ERROR. (thanks to Milan Zamazal)
2231 * fixed bug 191c: CLOS now does proper keyword argument checking as
2232 described in CLHS 7.6.5 and 7.6.5.1.
2233 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2234 longer raise an error; further, using a list as a for-as-arithmetic
2235 counter now raises a meaningful error.
2236 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2237 now understood by sequence creation functions such as MAKE-SEQUENCE
2239 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2240 a STRING-INPUT-STREAM.
2241 * compiler enhancement: SIGNUM is now better able to derive the type
2243 * type declarations inside WITH-SLOTS are checked. (reported by
2245 * fixed some bugs revealed by Paul Dietz' test suite:
2246 ** incorrect optimization of TRUNCATE for a positive first
2247 argument and negative second.
2248 ** compiler failure in let-convertion during flushing dead code.
2249 ** compiler failure while deriving type of TRUNCATE on an
2250 interval, containing 0.
2251 ** ASH of a negative bignum by a negative bignum count now returns
2253 ** intersection of CONS types now canonicalizes properly, fixing
2254 inconsistencies in SUBTYPEP.
2256 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2257 * incompatible change: The --disable-debugger command line
2258 option now clobbers the debugger at a more fundamental
2259 level, by redefining #'INVOKE-DEBUGGER instead of by
2260 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2261 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2262 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2263 enter the debugger REPL and then suffer recursive errors
2264 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2265 while under the new system, BREAK in --disable-debugger mode
2266 terminates the system just as an unhandled error would.
2267 * fixed compiler performance when processing loops with a step >1;
2268 * bug fix: DOCUMENTATION now retrieves generic function
2269 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2270 support has been systematized, and now supports the methods
2271 specified by ANSI, along with a default method and a method for
2272 slot documentation. (reported by Nathan Froyd)
2273 * bug fix: effective methods associated with a generic function are
2274 no longer cached over a change of that generic function's method
2275 combination. (reported by Andreas Fuchs)
2276 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2277 names a non-existent directory. (thanks to Andreas Fuchs)
2278 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2279 the PPC platform, lead to stack corruption; this has been fixed.
2280 (reported by Rainer Joswig)
2281 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2282 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2283 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2284 arithmetic in general, is now much more reliable on the PPC
2286 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2287 platform now returns the right answer.
2288 * optimization: restored some effective method precomputation in
2289 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2290 precomputation is now tunable.
2291 * optimization: compiler-internal data structure use has been
2292 reviewed, and changes have been made that should improve the
2293 performance of the compiler by about 20%.
2294 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2295 simple-base-strings and simple-bit-vectors is improved.
2296 * optimization: the optimization of 32-bit logical and arithmetic
2297 functions introduced in version 0.8.3 on the x86 has been
2298 implemented on the mips, ppc and sparc platforms; an
2299 implementation of the same facility, but for 64-bit arithmetic,
2300 has been added for the alpha.
2301 * microoptimization: the compiler is better able to make use of the
2302 x86 LEA instruction for multiplication by constants.
2303 * bug fix: in some situations compiler did not report usage of
2304 generic arithmetic in (SPEED 3) policy.
2305 * bug 145b fix: compiler used wrong type specifier while converting
2306 MEMBER-types to numeric.
2307 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2309 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2310 index argument now works properly on the Alpha platform.
2311 * bug fix: floating point exception treatment on the Alpha platform
2313 * bug fix: FILE-POSITION works much better on string input and
2314 output streams. (thanks to Nikodemus Siivola)
2315 * bug fix: many threading/garbage collection symptoms sorted.
2316 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2317 might be pseudo-atomic.
2318 * internal change: Stopping for GC is now done with signals not
2319 ptrace. GC is now done in whichever thread wanted it, instead of
2321 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2323 * bug fix: result form in DO is not contained in the implicit
2325 * incompatible change: ICR structure is changed; the value part of
2326 CONTINUATION is now called LVAR; corresponding functions are
2327 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2328 * added type deriver for ISQRT (thanks to Robert E. Brown).
2329 * bug fix: better support for loading from the command line when an
2330 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2332 * fixed some bugs revealed by Paul Dietz' test suite:
2333 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2334 ** ROUND and FROUND now give the right answer when given very
2335 small float arguments.
2336 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2338 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2339 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2340 ** LCM should return a non-negative integer.
2341 ** PARSE-INTEGER returned the index of a terminator instead of the
2342 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2343 ** PARSE-INTEGER returned an incorrect index being applied to a
2345 ** LCM with two arguments of 0 returns 0 rather than signalling
2347 ** unsigned addition of a 32-bit constant with the high bit set no
2348 longer causes an internal compiler error on the x86.
2349 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2350 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2351 with negative last argument.
2352 ** byte specifiers with zero size and position no longer cause
2353 an error during type derivation.
2354 ** bignum multiplication on the Alpha platform now returns the
2356 * porting: The system now builds on SuSE AMD64, although it still
2357 generates a 32-bit binary.
2358 * .fasl file incompatibility: The fasl file version number has
2359 been incremented (because of the changes to internal compiler
2360 data structures referred to above).
2362 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2363 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2364 more accurately, on the Darwin kernel running on PowerPC hardware.
2365 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2366 * Compiler code deletion notes now signal a condition of type
2367 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2368 an associated MUFFLE-WARNING restart.
2369 * The compiler now performs limited argument count validation of
2370 constant format strings in FORMAT, and where appropriate in ERROR,
2371 CERROR and WARN. (thanks to Gerd Moellmann)
2372 * New ASDF-INSTALL contrib can be used for automatic download and
2373 installation of third-party Lisp code from CCLAN or other sites
2375 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2376 which forces another thread to execute a function supplied by the
2378 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2379 now accept and act upon their :ELEMENT-TYPE keyword argument.
2380 (reported by Martin Atzmueller, Edi Weitz)
2381 * bug fix: FILE-POSITION now accepts position designators up to
2382 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2383 is the greater. (thanks to Patrik Nordebo)
2384 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2385 Kalle Olavi Niemitalo)
2386 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2387 printable. (reported by Eric Marsden)
2388 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2389 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2390 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2391 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2392 (thanks to Stig Sandoe)
2393 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2394 UNREAD-CHAR now correctly includes the unread character in the
2395 target sequence. (thanks to Gerd Moellmann)
2396 * bug fix in threaded builds: the system can now be suspended and
2397 resumed by shell job control with minimal disruption.
2398 * bug fixes in times and timezones >2038AD
2399 * better handling of "where is GNU make?" problem in build scripts
2400 (thanks to Nikodemus Siivola)
2401 * new optimization: inside a named function any reference to a
2402 function with the same name is considered to be a self-reference;
2403 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2404 optimization quality.
2405 * new optimization on x86: logical functions and + now have
2406 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2407 used when the result is truncated to 32 bits.
2408 * VALUES declaration is partially enabled.
2409 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2410 * fixed some bugs revealed by Paul Dietz' test suite:
2411 ** The system now obeys the constraint imposed by
2412 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2413 types form a lattice under type intersection.
2414 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2415 ** ASSOC now ignores NIL elements in an alist.
2416 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2417 and (1+ MOST-POSITIVE-FIXNUM) answers.
2418 ** The addition of a method with invalid qualifiers to a generic
2419 function does not cause an error to be signalled immediately;
2420 a warning is signalled, and the error is generated only on
2421 calling the generic function.
2422 * changed .fasl file version number, in part to add type codes for
2423 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2424 obscure ANSI requirements
2426 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2427 * fixed bug 148: failure to inline-expand a local function left
2428 garbage, confusing the compiler.
2429 * fixed bugs 3cd: structure slot readers perform type check if the
2430 slot can have an invalid value (i.e. it is either not initialized
2431 or can be written with a less specific slot writer).
2432 * bug fix: the compiler now traps array references to elements off
2433 the end of an array; previously, the bounds checking in some
2434 circumstances could go off-by-one.
2435 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2437 * type declarations for array element types now obey the description
2438 on the CLHS page "Declaration TYPE", as per discussions on
2439 sbcl-help around 2003-05-08. This means that a declaration
2440 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2441 declaration, all references to BAR will be asserted or assumed
2442 (with THE, so dependent on compiler policy) to involve objects of
2443 type FOO. Note that no such declaration is implied in
2444 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2445 * declared types of functions from the "Conditions"
2446 chapter. (reported by Paul Dietz)
2447 * bug fix: CERROR accepts a function as its first argument.
2448 * bug fix: NTH an NTHCDR accept a bignum as index
2449 arguments. (reported by Adam Warner)
2450 * optimization: character compare routines now optimize comparing
2451 against a constant character. (reported by Gilbert Baumann)
2452 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2453 argument now works properly on the MIPS platform.
2454 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2455 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2456 declared to have a functional type, does not cause an error
2458 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2459 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2460 * fixed bug: if last continuation of a deleted block has a
2461 destination, this destination should be deleted too. (reported by
2463 * fixed a bug in the bootstrap process: the host compiler's values
2464 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2465 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2466 test case from Patrik Nordebo)
2467 * improved the ability of the disassembler on the PPC platform to
2468 provide helpful disassembly notes.
2469 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2470 the class in more cases than previously.
2471 * bug fix: FILE-POSITION now understands :START and :END for
2472 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2473 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2474 8) data. (thanks to David Lichteblau)
2475 * bug fix: it is possible to add a method to a generic function
2476 without lambda list.
2477 * bug fix: reader failed to signal END-OF-FILE inside an
2478 object representation. (reported by Nikodemus Siivola)
2479 * fixed some bugs revealed by Paul Dietz' test suite:
2480 ** LAST and [N]BUTLAST should accept a bignum.
2481 ** condition slot accessors are methods.
2482 ** (VECTOR NIL) is a subtype of STRING.
2484 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2485 * minor incompatible change: some nonsensical specialized lambda
2486 lists (used in DEFMETHOD) which were previously ignored now signal
2488 * minor incompatible change: the system is now aware of the types of
2489 variables in the COMMON-LISP package, and will signal errors for
2490 most violations of these type constraints (where previously they
2491 were silently accepted).
2492 * minor incompatible change: COMPILE-FILE now uses the freedom
2493 afforded (ANSI 3.2.2.3) to use derived function types for
2494 functions defined in the same file. This also permits the system
2495 to warn on static type mismatches and function
2496 redefinition. (Currently it does not work with high DEBUG level.)
2497 * minor incompatible change: VALUES declaration is disabled.
2498 * When issuing notes, the compiler now signals a condition of type
2499 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2500 restart for use in user handlers. It is expected that the
2501 COMPILER-NOTE condition will eventually become a condition
2502 supertype to a hierarchy of note types, which will then be
2503 handleable in a similar fashion. However, at the moment, no such
2504 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2505 but it's an implementation detail, not a classification for the
2507 * Changes in type checking closed the following bugs:
2508 ** type checking of unused values (192b, 194d, 203);
2509 ** template selection based on unsafe type assertions (192c, 236);
2510 ** type checking in branches (194bc).
2511 * A short form of VALUES type specifier has ANSI meaning (it has
2512 increased the number of situations when SBCL cannot perform type
2514 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2515 keywords or constants is permissible.
2516 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2517 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2518 classes. (thanks to Antonio Martinez)
2519 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2520 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2521 argument to a pretty-print stream. Instead, it leaves any such
2522 operations to DESCRIBE-OBJECT methods.
2523 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2525 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2526 lambda lists. (thanks to David Lichteblau)
2527 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2528 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2529 (thanks to Gerd Moellmann)
2530 * fixed bug 246: increased compilation speed of long
2531 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2533 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2535 * DEFCONSTANT now throws a condition of type
2536 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2537 constant to a non-EQL value; CONTINUE and ABORT restarts
2538 respectively change and preserve the value.
2539 * fixed bug 63: The code walker, part of the implementation of CLOS,
2540 is now better at handling symbol macros.
2541 * bug fix: There is no longer an internal implementation type named
2542 CL:LENGTH. (reported by Raymond Toy)
2543 * bug fix: In macro-like defining macros/special operators the
2544 implicit block does not enclose the lambda list.
2545 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2546 enclosed in parentheses) are not suitable as type specifiers, and
2547 their use properly signals an error now.
2548 * bug fix: An argument count mismatch for a type specifier in code
2549 being compiled no longer causes an unhandled error at compile
2550 time, but signals a compile-time warning.
2551 * fixed simple vector readable printing
2552 * bug fix: DESCRIBE takes more care over whether the class
2553 precedence list slot of a class is bound before accessing it.
2554 (reported by Markus Krummenacker)
2555 * bug fix: FORMATTER can successfully compile pretty-printer format
2556 strings which use variants of the ~* directive inside.
2557 * bug fix: SEARCH now applies its TEST predicate to the elements of
2558 the arguments in the correct order. (thanks to Wolfhard Buss)
2559 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2560 in type inference. (thanks to Robert E. Brown)
2561 * bug fix: Reading in symbols with an explicit package name of ""
2562 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2563 * fixed some bugs revealed by Paul Dietz' test suite:
2564 ** NIL is now allowed as a structure slot name.
2565 ** Arbitrary numbers, not just REALs, are allowed in certain
2566 circumstances in LOOP for-as-arithmetic clauses.
2567 ** Multiple class redefinitions before slot access no longer
2568 causes a type error.
2569 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2570 association between the name and a class.
2571 ** Generic functions with non-standard method-combination and over
2572 five methods all of which return constants no longer return NIL
2573 after the first few invocations. (thanks to Gerd Moellmann)
2574 ** CALL-NEXT-METHOD with no arguments now passes the original
2575 values of the arguments, even in the presence of assignment.
2576 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2578 ** DISASSEMBLE works with closures and funcallable instances.
2579 ** ADD-METHOD now returns the generic function, not the new method.
2580 ** FIND-METHOD signals an error if the lengths of the specializers
2581 is incompatible with the generic function, even if the ERRORP
2583 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2584 which its argument is a member.
2585 ** DEFCLASS only redefines the class named by its class-name
2586 argument if that name is the proper name of the class;
2587 otherwise, it creates a new class.
2588 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2589 of the UNBOUND-SLOT condition to the name of the slot.
2590 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2591 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2592 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2593 SLOT-MAKUNBOUND in the specified fashion.
2595 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2596 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2597 cross-compilation host. As a consequence, we can now bootstrap our
2598 way up to SBCL starting with a bare gcc toolchain and human-readable
2599 source code (first the source to CLISP, then the source to SBCL).
2600 * A contributed module containing a partial implementation of the
2601 simple-streams interface has been included. (thanks to Rudi
2603 * A contributed module implementing the RFC1321 Message Digest
2604 Algorithm, known as MD5, has been included.
2605 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2606 no longer has any effect, as the code controlled by this feature
2607 has been deleted. (As far as we know, no-one has ever built using
2608 this feature, and its semantics were confused in any case).
2609 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2610 work on conditions (as required by the ANSI specification),
2611 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2612 expected behaviour on conditions. Users should note, however,
2613 that such behaviour is not required by the ANSI specification,
2614 and so use of this behaviour may render their code unportable.
2615 * fixed some bugs revealed by Paul Dietz' test suite:
2616 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2618 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2619 ** SXHASH on condition objects no longer returns NIL.
2620 ** :ALLOCATION :CLASS slots are better treated; their values are
2621 updated on class redefinition, and initforms inherited from
2622 superclasses are applied.
2623 ** REMOVE-METHOD returns its generic function argument even when
2624 no method was removed.
2625 ** SHARED-INITIALIZE now initializes the values of the requested
2626 slots, including those with :ALLOCATION :CLASS.
2627 ** ALLOCATE-INSTANCE now works on structure classes defined via
2628 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2630 ** SLOT-EXISTS-P now works on conditions, as well as structures
2632 ** MAKE-LOAD-FORM now has the required methods on
2633 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2634 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2635 keyword, and now implements the SLOT-NAMES argument.
2636 ** methods with &OPTIONAL arguments no longer allow too many
2637 arguments to be passed in the call without error.
2638 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2639 option is consistent with the required arguments of the generic
2640 function lambda list.
2641 * bug fix: REQUIRE accepts a string designator. (Thanks to
2643 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2644 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2645 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2646 * bug fix: The compiler now checks for duplicated variables in macro
2648 * bug fix: SETQ on globals returns the correct value.
2649 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2650 not a COMPILER-ERROR (followed by some other strange error on
2651 choosing the CONTINUE restart).
2652 * bug fix: make.sh and friends are now more consistent in the way that
2653 they look for GNU "make".
2655 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2656 * experimental native threads support (on x86 Linux >=2.4 only).
2657 This is not compiled in by default: you need to add :SB-THREAD to
2658 the target features. See the "Beyond ANSI" chapter of the manual
2660 * fix for longstanding nonANSIism: The old distinction between
2661 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2662 The return value from CL:FIND-CLASS is now a CLOS class, and
2663 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2664 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2665 * An interface to the MetaObject Protocol, as described in Kiczales,
2666 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2667 MIT Press, 1991, is available from the SB-MOP package.
2668 * incompatible change: the SB-PCL package should now be considered
2669 a private implementation detail, and no longer a semi-private MOP
2671 * minor incompatible change: due to rearrangement for threads, the
2672 control stack and binding stack are now allocated at arbitrary
2673 addresses instead of being hardcoded per-port. Users affected by
2674 this probably have to be doing advanced things with shared
2675 libraries, and will know who they are.
2676 * minor incompatible change: Previously, all --eval forms used were
2677 processed with READ before any of them were processed with EVAL.
2678 Now each --eval form is processed with both READ and EVAL before
2679 the next --eval form is processed. (Thus package operations like
2680 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2681 work as the user might reasonably expect.)
2682 * minor incompatible change: *STANDARD-INPUT* is now only an
2683 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2685 * minor incompatible change: Y-OR-N-P is now character-oriented,
2686 not line oriented. Also, YES-OR-NO-P now works without errors.
2687 (thanks to Antonio Martinez)
2688 * sb-aclrepl module improvements: an integrated inspector, added
2689 repl features, and a bug fix to :trace command.
2690 * Known functions, which cannot be open coded by the backend, are
2691 considered to be able to check types of their arguments. (fixing
2692 a bug report by Nathan J. Froyd)
2693 * fixed a bug in computing method discriminating functions: It is
2694 now possible to define methods specialized on classes which have
2695 forward-referenced superclasses. (thanks to Gerd Moellmann)
2696 * fixed evaluation order in optional entries (reported by Gilbert
2698 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2699 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2700 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2701 required-by-AMOP NAME argument, as well as CLASS and
2702 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2703 * fixed bug 20: DEFMETHOD can define methods using names that are
2704 not the proper names of classes to designate class specializers.
2705 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2706 Unix file descriptors, instead of blowing up. (thanks to Antonio
2708 * Garbage collection refactoring: user-visible change is that a
2709 call to the GC function during WITHOUT-GCING will not do garbage
2710 collection until the end of the WITHOUT-GCING. If you were doing
2711 this you were probably losing anyway.
2712 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2713 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2714 TYPEP the latter but not the former.
2715 * The compiler issues a full WARNING for calls to undefined functions
2716 with names from the CL package.
2717 * MAP-INTO for a vector destination is open coded. (reported by
2718 Brian Downing on c.l.l)
2719 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2720 documentation string.
2721 * fixed some bugs revealed by Paul Dietz' test suite:
2722 ** COPY-ALIST now signals an error if its argument is a dotted
2724 ** Condition slots are now accessed more correctly in the presence
2725 of multiple initargs for a given slot.
2726 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2727 exclude restarts of the same name associated with a different
2729 ** DEFCLASS of forward-referenced classes with another
2730 forward-referenced class in the superclasses list no longer
2732 ** Condition slots are now initialized once each, not multiple
2733 times. (thanks to Gerd Moellmann)
2734 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2736 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2737 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2738 arguments contain duplicated elements.
2739 ** RESTART-CASE understands local macros.
2740 ** RESTART-CASE associates exactly its own restarts with a condition.
2741 ** ENDP in safe mode checks its argument to be of type LIST.
2742 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2743 in question is unbound.
2744 ** Optimized MAKE-INSTANCE functions no longer cause internal
2745 assertion failures in the presence of duplicate initargs.
2746 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2747 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2748 propagated even in the ctor optimized implementation of
2750 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2752 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2753 * a better implementation of SXHASH on (simple) bit vectors,
2754 measured both in execution speed and in distribution of results
2755 over the positive fixnums, has been installed. Likewise, a better
2756 implementation of EQUAL for simple bit vectors is now available.
2757 * fixed CEILING optimization for a divisor of form 2^k.
2758 * fixed bug 240 (emitting extra style warnings "using the lexical
2759 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2760 by Antonio Martinez)
2761 * fixed SXHASH, giving different results for NIL depending on type
2762 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2763 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2764 variable. (found by Rolf Wester)
2765 * SBCL does not ignore type declarations for special
2766 variables. (reported by rif on c.l.l 2003-03-05)
2767 * some bug fixes in contrib/sb-aclrepl/
2768 * fixed some bugs revealed by Paul Dietz' test suite:
2769 ** a bug in the CONS type specifier, whereby the CAR and CDR
2770 types got intertwined, has been fixed;
2771 ** the type system is now able to reason about the interaction
2772 between INTEGER and RATIO types more completely;
2773 ** APPEND, [N]REVERSE and NRECONC check that those their
2774 arguments, which must be proper lists, are really so;
2775 ** An array specialized to be unable to hold elements has been
2776 implemented, as required -- yes, really -- by ANSI;
2777 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2778 on malformed property lists;
2780 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2781 * incompatible packaging change: in line with Unix convention,
2782 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2783 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2784 * REQUIRE and PROVIDE are now optionally capable of doing something
2785 useful. See the documentation string for REQUIRE.
2786 * infrastructure for a managed SBCL contrib system: contributed
2787 modules in this release include:
2788 ** the ASDF system definition facility;
2789 ** an interface to the BSD Sockets API;
2790 ** an ACL-like convenience interface to the repl;
2791 (thanks to Kevin Rosenberg)
2792 ** an implementation of ROTATE-BYTE, with efficient implementation
2794 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2795 gives a closer approximation to the right answer than previously.
2796 (thanks to Raymond Toy)
2797 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2798 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2799 optional environment arguments, as required by ANSI.
2800 * fixed bugs in other functions taking environment objects, allowing
2801 calls with an explicit NIL environment argument to be compiled
2803 * fixed bug 228: primary return values from
2804 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2805 COMPILE or FUNCTION.
2806 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2807 :TYPE will no longer signal errors on innocuous objects.
2808 * fixed bug 231b: SETQ is better at respecting type declarations in
2809 the lexical environment.
2810 * fixed a bug in DEFCLASS: classes named by symbols with no or
2811 unprintable packages can now be defined.
2812 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2813 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2814 * fixed compiler failure related to checking types of functions.
2815 (reported by Robert E. Brown)
2816 * the compiler is now much more consistent in its error-checking
2817 treatment of bounding index arguments to sequence functions: in
2818 (SAFETY 3) code, errors will be signalled in almost all cases if
2819 invalid sequence bounding indices are passed to functions defined
2820 by ANSI to operate on sequences.
2821 * fixed a bug in the build procedure: documentation of SBCL-specific
2822 packages is now preserved and available in the final Lisp image.
2823 * lifted FDEFINITION lookup out of loops in the implementation of
2824 many list operations. (thanks to Robert E. Brown)
2825 * fixed a bug in the reader: the #n# reader macro now works for
2826 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2827 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2828 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2829 E. Brown and Tony Martinez respectively)
2830 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2831 SLOT-BOUNDP: in particular, it is now able to optimize them much
2832 better, and is now not vulnerable to having packages renamed.
2833 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2834 always called when a slot is not present in an instance. (thanks
2836 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2837 longer signal an error on unreading a character following EOF on
2838 the previous constituent stream. (thanks to Tony Martinez)
2839 * fixed some bugs revealed by Paul Dietz' test suite:
2840 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2841 not just nonnegative fixnums;
2842 ** the logical bit-array operators such as BIT-AND now accept an
2843 explicit NIL for their "opt-arg" argument (to indicate a
2844 freshly-consed result bit-array);
2845 ** ELT now signals an error on an invalid sequence index in safe
2847 ** the type system is now cleverer about negations of numeric
2848 types, and consequently understands the BIGNUM and RATIO types
2850 ** the type system is now cleverer about the interaction between
2851 INTEGER and RATIO types: while bugs still remain, many more
2852 cases are accurately computed;
2853 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2854 if it is in the last clause;
2855 ** CONSTANTLY now correctly returns a side-effect-free function in
2857 ** DECLARE is no longer treated as a special-operator; in
2858 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2859 * incremented fasl file version number due to the change in the
2860 DEFSTRUCT-SLOT-DESCRIPTION structure.
2862 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2863 * minor incompatible change: code processed by the "interpreter" or
2864 EVAL now has a compilation optimization policy of (DEBUG 2)
2865 (changed from (DEBUG 1)) to improve debuggability of interactive
2866 development, and to allow the use of the debug RETURN command in
2868 * an experimental implementation of the RETURN command for the
2869 debugger has been included. (thanks to Frederik Kuivinen)
2870 * fixed bug 62: constraints were not propagated into a loop.
2871 * fixed bug in embedded calls of SORT (reported and investigated by
2873 * fixed some bugs revealed by Paul Dietz' test suite:
2874 ** printing and reading of arrays with some dimensions having
2875 length 0 (thanks to Gerd Moellmann);
2876 ** BOA constructor with &AUX argument without a default value does
2877 not cause a type error;
2878 ** CONSTANTP now returns true for all self-evaluating objects.
2880 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2881 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2882 accessors that are related by inheritance, as specified in the
2883 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2884 Valtteri Vuorikoski)
2885 * The compiler is now able to inline functions that were defined in
2886 a complex lexical environment (e.g. inside a MACROLET).
2887 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2889 * Support for the upcoming FreeBSD-5.0 release has been included.
2890 (thanks to Dag-Erling Smorgrav)
2891 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2892 effect when it is not in a toplevel context.
2893 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2894 stand a better chance of being correct. (thanks to Gerd
2896 * fixed bug in COERCE, which now signals an error on coercing a
2897 rational to a bounded real type which excludes the expected
2899 * The compiler is now able to derive types more accurately from the
2900 COERCE and COMPILE functions.
2901 * fixed bug 223: functional binding is considered to be constant
2902 only for symbols in the CL package.
2903 * fixed bug 231: SETQ did not check the type of a variable being set
2904 (reported by Robert E. Brown)
2905 * A new optimization for MAKE-INSTANCE has been included, fixing
2906 various bugs (including relating to :ALLOCATION :CLASS slots and
2907 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2909 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2910 ** As required by ANSI, LOOP now disallows anonymous collection
2911 clauses such as COLLECT I in conjunction with aggregate boolean
2912 clauses such as THEREIS (= I 1);
2913 ** LOOP now signals an error when any variable is reused in the
2914 same loop (including the potentially useful construct analogous
2915 to WITH A = 1 WITH A = (1+ A);
2916 ** IT is only a special loop symbol within the first clause of a
2917 conditional loop clause;
2918 ** LOOP with a typed iteration variable over a hashtable now
2919 signals a type error iff it should.
2920 * fixed some other bugs revealed by Paul Dietz' test suite:
2921 ** FILE-STREAM now names the class previously known as FD-STREAM;
2922 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2923 argument) no longer signals an error;
2924 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2925 of the slot symbol, rather than using the current package
2926 ((:CONC-NAME "") continues to intern the slot's name in the
2928 * incremented fasl file version number, because of the incompatible
2929 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2930 of the new implementation of DEFINE-COMPILER-MACRO.
2932 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2933 * Support for building SBCL for MIPS platforms running in
2934 little-endian mode has now been checked in, and basic
2935 functionality on said platforms verified.
2936 * minor incompatible change: PCL now records the pathname of a file
2937 in which methods and the like are defined, rather than its
2939 * minor incompatible change: TRUENAME now considers the truename of
2940 a file naming a directory to be the pathname with :DIRECTORY
2941 component indicating that directory.
2942 * minor incompatible change: a NAMED clause in the extended form of
2943 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2944 reason for the previous behaviour is unclear.
2945 * more systematization and improvement of CLOS and MOP conformance
2946 in PCL (thanks to Gerd Moellman and Pierre Mai):
2947 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2949 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2950 primary methods with no specializers;
2951 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2953 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2954 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2955 FORWARD-REFERENCED-CLASSes; error reporting on
2956 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2958 ** SXHASH on CLOS instances now uses a slot internal to the
2959 instance to return different numbers on distinct instances,
2960 while preserving the same return value through invocations of
2962 ** DEFMETHOD signals errors when methods with longer incongruent
2963 lambda lists are added to generic functions;
2964 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2965 CLASS, as specified in AMOP;
2966 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2967 on the order returned by the primary method for classes of
2968 class STANDARD-CLASS;
2969 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2970 * fixed some bugs shown by Paul Dietz' test suite:
2971 ** DOLIST puts its body in TAGBODY;
2972 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2974 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2975 value producing form;
2976 ** if more variables are given to PROGV than values, extra
2977 variables are bound and made to have no value;
2978 ** NSUBSTITUTE on list arguments gets the right answer with
2980 ** ELT signals an error of type TYPE-ERROR when the index argument
2981 is not a valid sequence index;
2982 ** LOOP signals (at macroexpansion time) an error of type
2983 PROGRAM-ERROR when duplicate variable names are found;
2984 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2985 ** FUNCALL of special-operators now cause an error of type
2987 ** PSETQ now works as required in the presence of side-effecting
2988 symbol-macro places;
2989 ** NCONC accepts any object as its last argument;
2990 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2992 ** loop-for-as-package does not require a package to be explicitely
2994 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2996 * fixed bug 166: compiler preserves "there is a way to go"
2997 invariant when deleting code.
2998 * fixed bug 172: macro lambda lists with required arguments after
2999 &REST arguments now cause an error to be signalled. (thanks to
3001 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3002 bug. (thanks to Matthew Danish)
3003 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3005 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3006 arguments are lexically rebound. (thanks to Gerd Moellmann and
3008 * fixed bug 194: error messages are now more informative when there
3009 is no primary method applicable in a call to a generic
3010 function. (thanks to Gerd Moellmann)
3011 * fixed bug in command line argument checking (thanks to Julian
3013 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3014 (thanks to Matthew Danish)
3015 * incremented fasl file version number, because of the
3016 SXHASH-related changes in the layout of CLOS data structures
3018 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3019 * minor incompatible change: The runtime (the Unix executable named
3020 "sbcl") is now much pickier about the .core files it will load.
3021 Essentially it now requires .core files to descend from the same
3022 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3023 as the runtime does. (The intent is to prevent the crashes which
3024 can occur, and which can even be reported as mysterious failures,
3025 when people patch the sources or change the build parameters
3026 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3027 sbcl and .core files.)
3028 * fixed bug: VALUES-LIST is no longer optimized away.
3029 * fixed bug 142: The FFI conversion of C string values to Lisp
3030 string values no longer conses excessively. (thanks to Nathan
3031 Froyd porting Raymond Toy's fix to CMU CL)
3032 * began to systematize and improve MOP conformance in PCL (thanks to
3033 Nathan Froyd, Gerd Moellman and Pierre Mai):
3034 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3036 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3037 ** FINALIZE-INHERITANCE is now called on class finalization;
3038 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3039 argument precedence order.
3040 * fixed bug 202: The compiler no longer fails on functions whose
3041 derived types contradict their declared type.
3042 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3043 so it can be non-toplevel.
3044 * The fasl file version number has changed (because of the new
3045 implementation of DEFMACRO).
3046 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3047 safe code, that any length requirement by their type-specifier
3048 argument is valid. The exceptions to this are described in bug
3050 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3051 functions now check the types of their inputs as required by ANSI.
3052 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3053 introduced symbol is DECLAREd to be SPECIAL.
3054 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3055 * fixed bug: :COUNT argument to sequence functions may be negative
3056 * fixed bug: body of DO-SYMBOLS may contain declarations
3057 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3058 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3059 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3060 symbol macro only once
3061 * fixed printing of call frame when argument list is unavailable
3062 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3063 * compiler no longer signals WARNING on unknown keyword
3066 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3067 * A beta-quality port to the mips architecture running Linux,
3068 based on the old CMUCL backend, has been made. It has been tested
3069 on a big-endian kernel, and works sufficiently well to be able to
3070 rebuild itself; it has not been tested in little-endian mode.
3071 * fixed an inconsistency between gencgc.c and purify.c which made
3072 dumping/loading .core files unreliable
3073 * fixed bug 120a: The compiler now deals correctly with IFs where
3074 the consequent is the same as the alternative, instead of
3075 misderiving the return type. (thanks to Alexey Dejneka)
3076 * fixed bug 113: Logical pathnames are now dumpable (the logical
3077 host is resolved at load-time, throwing an error if it is not
3079 * fixed bug 174: FORMAT's error message is slightly clearer when a
3080 non-printing character is used in a format directive.
3081 * fixed several bugs in compiler checking of type declarations, i.e.
3082 violations of the Python "declarations are assertions" principle
3083 (thanks to Alexey Dejneka)
3084 * fixed several bugs in PCL's error checking (thanks to Gerd
3086 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3088 * fixed bug in compilation of functions as first class values
3089 (thanks to Antonio Martinez-Shotton)
3090 * The compiler's handling TYPE-ERRORs which it can prove will
3091 inevitably happen at runtime has been cleaned up and corrected
3092 in several ways. (thanks to Alexey Dejneka)
3093 * improved argument type checking for various basic arithmetic
3094 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3095 lavished on them in the past that they can be compiled in many
3096 ways in different special cases
3097 * fixed bug 181: compiler checks validity of user supplied type
3099 * cleaned up code flushing in optimization: Function calls which
3100 should signal errors for safety purposes (e.g. which ANSI says
3101 should signal errors when their arguments are of incorrect type)
3102 are no longer optimized away.
3103 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3104 * incremented fasl file version number, because changes in the
3105 implementation of sequence functions like COERCE caused
3106 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3107 in old inline expansions) to become undefined. (Actually these
3108 changes were later undone, so we might very well be binary
3109 compatible with 0.7.7 after all, but leaving the version number
3110 incremented seemed like the simplest and most conservative
3113 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3114 * An alpha-quality port to the parisc architecture running Linux,
3115 based on the old CMUCL backend, has been made. This, even more so
3116 than the other backends, should be considered still a work in
3117 progress; known problems include that the Linux kernel in 64-bit
3118 mode does not propagate the correct sigcontext structure to
3119 userspace, and consequently SBCL on a parisc64 kernel will not
3121 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3122 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3123 suppressed.) Also now that the compiler is looking at declarations
3124 in the environment, it checks optimization declarations as well,
3125 and suppresses inlining when (> DEBUG SPEED).
3126 * More fixes have been made to treatment of floating point exception
3127 treatment and other Unix signals. In particular, floating point
3128 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3129 * The detection and handling of control stack exhaustion (infinite
3130 or very deeply nested recursion) has changed. Stack exhaustion
3131 detection is now done by write-protecting pages at the OS level
3132 and applies at all optimization settings; when found, a
3133 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3134 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3135 be caught using IGNORE-ERRORS.
3136 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3137 correctly when transforming recursive LABELS functions to LETs.
3138 (thanks to Alexey Dejneka)
3139 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3140 that are names of constants or global variables.
3141 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3142 alien routines with docstrings.
3143 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3144 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3146 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3147 (thanks to Eric Marsden)
3148 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3149 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3150 * Bug fix: The SPARC backend can now compile functions involving
3151 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3152 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3153 to a functions expecting a PATHNAME-DESIGNATOR.
3154 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3155 lambda lists. (thanks to Alexey Dejneka)
3156 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3157 functions so that EXPORTing the name of the class doesn't cause
3158 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3159 to Antonio Martinez for reporting this)
3160 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3161 constant-folded without failing an assertion. (thanks to Einar
3162 Floystad Dorum for reporting this)
3163 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3164 types are now dealt with more correctly by the compiler.
3165 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3166 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3167 behaviour analogous to RENAME-FILE. This puts its behaviour more
3168 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3170 * The fasl file version number has changed again. (because of the
3171 bug fix involving the names of PCL MAKE-INSTANCE functions)
3173 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3174 * bug fix: Floating point exceptions are treated much more
3175 consistently on the x86/Linux and PPC/Linux platforms.
3176 * Array initialization with :INITIAL-ELEMENT is now much faster for
3177 cases when the compiler cannot open code the array creation, but
3178 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3179 array accesses have also seen a speed increase.
3180 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3181 specified. (This was at the root of some bad interactions between
3182 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3184 * bug fix: Internal error arguments for undefined functions are now
3185 computed correctly on the PPC/Linux platform.
3186 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3187 Raymond Toy's patch for CMU CL)
3188 * Support for the Solaris 9 operating environment has been included
3189 (thanks to Daniel Merritt)
3190 * A very ugly but hopefully complete draft of the missing FFI chapter
3191 of the manual has been created by reformatting the corresponding
3192 CMU CL manual chapter into (currently very ugly and incoherent)
3193 DocBook and bringing it up to date for SBCL behavior. Thus, the
3194 manual is now essentially complete, at least by my extreme
3195 once-and-only-once standards, whereby it's acceptable to refer to
3196 the doc strings of SB-EXT functions as the primary documentation.
3197 * The fasl file version number has changed again, due to cleanup of
3198 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3199 is no longer a static symbol.)
3201 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3202 * SBCL now builds with OpenMCL (version 0.12) as the
3203 cross-compilation host; also, more progress has been made toward
3204 bootstrapping under CLISP.
3205 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3207 * bug 158 fixed: The compiler can now deal with integer loop
3208 increments different from 1; fixing this turned out also to fix
3210 * bug 169 fixed: no more bogus warnings about using lexical bindings
3211 despite the presence of perfectly good SPECIAL declarations (thanks
3212 to David Lichteblau)
3213 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3214 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3215 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3217 * bug 180 fixed: Method combination specifications no longer ignore
3218 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3219 * bug fix: Structure type predicate functions now check their argument
3220 count as they should.
3221 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3222 correctly. (thanks to Pierre Mai)
3223 * minor incompatible change: The --noprogrammer option is deprecated
3224 in favor of the new --disable-debugger option, which is very similar.
3225 (The major difference is that it takes effect at a slightly different
3226 time at startup, causing handling of errors in --sysinit and
3227 --userinit files will be affected differently.) The
3228 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3229 been added to allow this functionality to be controlled from ordinary
3230 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3231 maintainers, who might want to run non-interactive scripts to
3232 build SBCL cores which will later be used interactively.)
3233 * minor incompatible change: The LOAD function no longer, when given
3234 a wild pathname to load, loads all files matching that pathname.
3235 Instead, an error of type FILE-ERROR is signalled.
3237 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3238 * bug 147 fixed: The compiler preserves its block link/count
3239 invariants more correctly now so that it doesn't crash. (thanks
3241 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3243 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3244 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3245 no-value-for-this-character as NIL (as specified by ANSI).
3246 * HOST-NAMESTRING on physical pathnames now returns a string that is
3247 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3248 (thanks to Christophe Rhodes)
3249 * The Alpha port handles icache flushing more correctly. (thanks to
3251 * More progress has been made toward bootstrapping under CLISP. (thanks
3252 to Christophe Rhodes)
3253 * The fasl file format has changed again, because dynamic loading
3254 on OpenBSD (which has non-ELF object files) motivated some cleanups
3255 in the way that foreign symbols are transformed and passed around.
3256 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3257 is no longer treated as whitespace by the reader, but instead as
3258 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3259 |A
\7fB|, instead of A as it used to.
3261 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3262 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3263 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3264 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3265 this as of 0.7.1.45, but was left out of the previous news section
3266 (thanks to Dan Barlow)
3267 * SBCL now runs on the Solaris operating system on SPARC architectures
3268 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3269 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3270 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3271 * SPARC backend cleanups, allowing builds of cores optimized for V8
3272 and V9 SPARCS, and also emission of code targeted to a particular
3273 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3275 * SBCL is closer to bootstrapping under CLISP, thanks to various
3276 fixes by Christophe Rhodes.
3277 * The fasl file format has changed again, to allow the compiler's
3278 INFO database to support symbol macros.
3279 * The user manual (in doc/) is formatted into HTML more nicely.
3280 (thanks to coreythomas)
3281 * The system is smarter about SUBTYPEP relationships, especially
3282 those involving NOT types (including types such as ATOM which are
3283 represented internally using NOT types). Thus SUBTYPEP is less
3284 likely to return (VALUES NIL NIL) in general, and in particular
3285 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3287 * The fasl file format has changed again, because the internal
3288 representation of types now includes a new slot to support the new
3289 SUBTYPEP-of-NOT-types logic.
3290 * (not a change in the main branch of SBCL, but a related prototype
3291 which can hopefully be merged into the main branch of SBCL in the
3292 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3293 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3294 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3295 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3296 dispatch macro character combinations. (thanks to Alexey Dejneka)
3297 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3298 that superficially look like logical namestrings correctly.
3299 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3301 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3302 default supplied-p) for &optional and &key arguments. (thanks to
3305 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3306 * incompatible change: The compiler is now less aggressive about
3307 tail call optimization, doing it only when (> SPACE DEBUG) or
3308 (> SPEED DEBUG). (This is an incompatible change because there are
3309 programs which relied on the old CMU-CL-style behavior to optimize
3310 away their unbounded recursion which will now die of stack overflow.)
3311 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3312 for non-GENCGC systems has been increased to 20M (since that
3313 seems much closer to the likely performance optimum for modern
3314 systems than the old 4M value was)
3315 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3316 and *DEBUG-PRINT-LEVEL*
3317 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3318 of CMU CL's support for SPARC, and various endianness and other
3319 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3320 * new syntactic sugar for the Unix command line: --load foo.bar is now
3321 an alternate notation for --eval '(load "foo.bar")'.
3323 ** The system now detects stack overflow and handles it gracefully,
3324 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3325 optimization settings. (This is a good thing in general, and
3326 its introduction in this version should be particularly timely
3327 for anyone whose code fails because of suppression of tail
3329 ** The system now hunts for the C variable "environ" in a more
3330 devious way, to avoid segfaults when the C library version differs
3331 between compile time and run time. (thanks to Christophe Rhodes)
3332 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3333 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3334 ** The compiler no longer issues bogus style warnings for undefined
3335 classes in the same source file as the DEFCLASSes which defined
3336 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3338 ** fixes in CONDITION class precedence list for undefined function
3339 errors (thanks to Alexei Dejneka)
3340 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3341 correctly. (thanks to Dan Barlow)
3342 ** portability fixes aiming at bootstrapping under CLISP (thanks
3343 to Dave McDonald and Christophe Rhodes)
3344 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3345 ** fixes in type translation and and type inference (thanks to
3347 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3348 ** various fixes in the new ports (thanks to Dan Barlow)
3349 * several changes related to debugging:
3350 ** suppression of tail recursion, as noted above
3351 ** stack overflow detection, as noted above
3352 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3353 is now the default. (For some time encapsulation has been more
3354 reliable than the breakpoint-based :ENCAPSULATE NIL
3355 implementation, at least on X86 systems; and I just noticed that
3356 encapsulation also seems closer to the spirit of the ANSI
3359 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3361 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3362 up properly again. (There was a packaging bug in 0.7.0 which
3363 left their definitions in SB-SYS::LOAD-FOREIGN and
3364 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3365 vital for most things which interface to C-level interfaces,
3366 like extensions working with sockets or databases or
3367 Perl-compatible regexes or whatever, and the need to fix
3368 this bug is the main reason that 0.7.1 was released so
3370 ** DEFGENERIC is now choosier about the methods it redefines, so that
3371 reLOADing a previously-LOADed file containing DEFGENERICs does
3372 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3373 cycle now works as it should. (thanks to Alexey Dejneka)
3374 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3375 fixed by Christophe Rhodes. (He actually submitted this patch
3376 months ago, and I delayed until after 0.7.0.)
3377 ** Bug 111 (internal compiler confusion about runtime checks on
3378 FUNCTION types) was fixed by Alexey Dejneka.
3379 * Some internal cleanups (getting rid of variables which aren't
3380 needed now that the byte interpreter is gone) caused the fasl
3381 file format number to change again.
3383 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3384 * major incompatible change: The default fasl file extension, i.e. the
3385 default extension for files produced by COMPILE-FILE, has changed
3386 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3388 ** There are many changes in the implementation of the compiler.
3389 SBCL is now essentially a compiler-only implementation of ANSI
3390 Common Lisp. EVAL still "interprets" a few special cases, but
3391 almost all the interesting cases are handled by creating
3392 a LAMBDA expression, calling COMPILE on it, then calling
3393 FUNCALL on the result.
3394 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3395 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3396 Since the code is newer, there might still be some new bugs
3397 (though not as many as before Martin Atzmueller's fixes:-). But
3398 the new code is substantially simpler and clearer, and hopefully
3399 any remaining bugs will be simpler, less fundamental, and more
3400 fixable then the bugs in the old code.
3401 ** The revised compiler is still a little unsteady on its feet.
3403 *** The debugging information it produces (particularly the names
3404 of FUNCTION objects) is sometimes much less useful than what
3405 the old compiler produced.
3406 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3407 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3408 within a MACROLET) has been temporarily weakened.
3409 ** There are new compiler optimizations for various functions:
3410 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3411 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3412 *** the math functions TRUNCATE, FLOOR, and CEILING
3413 *** the function-of-all-trades COERCE
3414 Mostly these should be transparent, but there's one
3415 potentially-annoying problem (bug 117): when the compiler
3416 inline-expands a function and does type analysis on the result,
3417 it can create control paths which have type mismatches, and
3418 when it can't prove that those control paths aren't taken,
3419 it will issue WARNINGs about the type mismatches. This is
3420 a particular problem in practice for the new sequence functions.
3421 It's not clear how this should be fixed, and for now, a
3422 workaround is given in the entry for 117 in the BUGS file.
3423 ** (Because of the interaction between the two previous items --
3424 occasional inlining problems and new inline expansions -- some
3425 of the new sequence function optimizations won't really kick in
3426 completely until debugging information, and then inlining, are
3427 straightened out in some future version.)
3428 * minor incompatible changes:
3429 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3430 DIRECTORY no longer implicitly promotes NIL slots of its
3431 pathname argument to :WILD. In particular, when you ask for the
3432 contents of a directory (which you used to be able to do without
3433 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3434 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3435 ** changes in behavior that ANSI explicitly defines to be
3436 implementation dependent:
3437 *** The new compiler-only implementation still conforms with ANSI,
3438 but acts a little different than before. Besides the obvious
3439 changes in performance tradeoffs (that the cost per form passed
3440 to EVAL has gone up, and the cost per form executed by EVAL
3441 has gone down), the behavior of the system changes a little
3442 because there are no longer any interpreted function objects.
3443 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3444 e.g. doing COMPILE on the output of interactive DEFUN is
3446 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3448 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3449 has changed from (:CL) to NIL.
3450 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3451 e.g. #\Newline and #\Tab, don't have names specified in the
3452 ANSI Common Lisp standard, is now based on their ASCII symbolic
3453 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3454 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3455 are no longer used for output.
3456 ** changes in internal implementation constants:
3457 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3458 4 million. (If your application spends a lot of time GCing and
3459 you have a lot of RAM, you might want to experiment with
3460 increasing it even more.)
3461 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3462 However, almost all old code should still continue to work without
3463 immediate update, as SB-C-CALL is now a (deprecated) nickname
3465 ** Old operator names in the style DEF-FOO are now deprecated in
3466 favor of new corresponding names DEFINE-FOO, for consistency with
3467 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3468 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3469 internal symbols, but a few supported extensions like
3470 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3471 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3472 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3473 etc. as you get deeper into recursive calls to the debugger
3474 command loop, instead of the old "5]", "5]]", "5]]]"
3475 sequence. (I was motivated to do this when squabbles between
3476 ILISP and SBCL left me very deeply nested in the debugger. In the
3477 short term, this change will probably provoke more ILISP/SBCL
3478 squabbles, but hopefully it will be an improvement in the long run.)
3479 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3480 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3481 function, since declaiming return types involving aliens is
3482 (1) annoyingly messy to do by hand and (2) vital to efficient
3483 compilation of code which calls such functions.
3484 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3485 longer reexported by the SB-EXT package. They're solely useful
3486 for alien code, so it seems more logical that you should get
3487 them from the SB-ALIEN package, not in SB-EXT.
3488 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3489 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3490 features. Instead, the code that they used to control is always
3491 built into the system.
3492 * many other bug fixes
3493 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3494 advantage of the new EVAL-WHEN stuff and to clean them up in
3495 general, and they are now more ANSI-compliant in a number of
3496 ways. Martin Atzmueller is responsible for a lot of this.
3497 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3499 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3500 *** correct ERROR type for various file operations
3501 *** some fixes for Lisp streams
3502 *** DEFMETHOD syntax checking
3503 *** changing old weird representation of debug information as
3504 strings (which, among their other deficiencies, don't transform
3505 correctly when you rename packages, and don't change their
3506 print representation when you change things like *PACKAGE*
3507 and *PRINT-LENGTH*) to symbols and lists of symbols
3508 He also made several improvements and fixed several bugs in DESCRIBE.
3509 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3511 *** misbehavior of WRITE-STRING/WRITE-LINE
3512 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3513 and several other LOOP problems as well
3514 *** DIRECTORY when similar filenames are present
3515 *** DEFGENERIC with :METHOD options
3516 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3517 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3518 *** argument ordering in FIND with :TEST option
3519 *** mishandled package designator argument in APROPOS-LIST
3520 *** various problems in the backquote readmacro
3521 *** a bug in APROPOS
3522 *** probably some others that I'm not describing very well here,
3523 since the CVS log documents them by reference to sbcl-devel
3524 messages, and the SourceForge archives aren't working well.:-(
3525 ** Dan Barlow improved the Alpha port (and is making progress on the
3526 PPC port, for those of you who think different).
3527 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3528 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3529 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3530 continues to work on the SPARC port (for those of you in a position
3531 to look down upon our little PC-compatible boxes from a great height).
3532 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3533 and uses that copy, rather than the current dynamic value, when
3534 it's trying to decide whether to truncate output. Thus e.g.
3535 (let ((*print-lines* 50))
3536 (pprint-logical-block (stream nil)
3538 (let ((*print-lines* 8))
3539 (print (aref possiblybigthings i) stream)))))
3540 should now truncate the logical block only at 50 lines, instead of
3541 often truncating it at 8 lines, as it did before.
3542 * The doc/cmucl/ directory, containing old CMU CL documentation
3543 from the time of the fork, is no longer part of the base system.
3544 SourceForge has shut down its anonymous FTP service, and with it
3545 my original plan for distributing the old CMU CL documentation
3546 there. For now, if you need these files you can download an old
3547 SBCL source release and extract them from it.
3548 * The fasl file version number changed again, for dozens of reasons,
3549 some of which are apparent above.
3551 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3552 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3553 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3554 patch, so that DISASSEMBLE output is much nicer.
3555 * The code in the SB-PROFILE package now seems reasonably stable.
3556 I still haven't decided what the final interface should look like
3557 (I'd like PROFILE to interact cleanly with TRACE, since both
3558 facilities use function encapsulation) but if you have a need
3559 for profiling now, you can probably use it successfully with
3560 the current CMU-CL-style interface.
3561 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3562 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3563 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3564 link now returns the dangling link itself, and for similar
3565 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3566 link itself. (In these cases the old code signalled an error and
3567 looped endlessly, respectively.) Thus, DIRECTORY now works even
3568 in the presence of dangling and cyclic symbolic links.
3569 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3570 is now a supported extension again, since the consensus on
3571 sbcl-devel was that it can be useful for ordinary development
3572 work, not just for debugging SBCL itself.
3573 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3574 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3575 that currently-defined functions might be redefined later with
3576 different return types.
3577 * Hash tables can be printed readably, as inspired by CMU CL code
3578 of Eric Marsden and SBCL code of Martin Atzmueller.
3579 * better error handling in CLOS method combination, thanks to
3580 Martin Atzmueller porting Pierre Mai's CMU CL patches
3581 * more overflow fixes for >16Mbyte I/O buffers
3582 * A bug in READ has been fixed, so that now a single Ctrl-D
3583 character suffices to cause end-of-file on character streams.
3584 In particular, now you only need one Ctrl-D at the command
3585 line (not two) to exit SBCL.
3586 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3588 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3589 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3590 does the right thing.
3591 * The implementation of some type tests, especially for CONDITION
3592 types, is now tidier and maybe faster, due to CMU CL code
3593 originally by Douglas Crosher, ported by Martin Atzmueller.
3594 * Some math functions have been fixed, and there are new
3595 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3596 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3597 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3598 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3599 likely to be built into sbcl-0.7.0.)
3600 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3601 CL, ported to SBCL by Martin Atzmueller)
3602 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3603 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3604 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3605 * There's a new slam.sh hack to shorten the edit/compile/debug
3606 cycle for low-level changes to SBCL itself, and a new
3607 :SB-AFTER-XC-CORE target feature to control the generation of
3608 the after-xc.core file needed by slam.sh.
3609 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3610 COMPILE-FILE is no longer supported, so that now every function
3611 gets an entry point, so that block compilation looks a little
3612 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3613 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3614 returns the number of bytes consed since the system started,
3615 rather than the number consed since the first time the function
3616 was called. (The new definition parallels ANSI functions like
3617 CL:GET-INTERNAL-RUN-TIME.)
3618 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3619 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3620 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3621 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3622 * Fasl file version numbers are now independent of the target CPU,
3623 since historically most system changes which required version
3624 number changes have affected all CPUs equally. Similarly,
3625 the byte fasl file version is now equal to the ordinary
3628 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3629 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3630 is no longer recognized. I apologize for this, because it was
3631 listed in SB-EXT as a supported extension, but I found that
3632 its existing behavior was poorly specified, as well as incorrectly
3633 specified, and it looked like too much of a mess to straighten it
3634 out. I have enough on my hands trying to get ANSI stuff to work..
3635 * many patches ported from CMU CL by Martin Atzmueller, with
3636 half a dozen bug fixes in pretty-printing and the debugger, and
3637 half a dozen others elsewhere
3638 * fixed bug 13: Floating point infinities are now supported again.
3639 They might still be a little bit flaky, but thanks to bug reports
3640 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3641 as flaky as they were.
3642 * The --noprogrammer command line option is now supported. (Its
3643 behavior is slightly different in detail from what the old man
3644 page claimed it would do, but it's still appropriate under the
3645 same circumstances that the man page talks about.)
3646 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3647 are now supported, and enabled by default. Thus, the compiler can
3648 handle many floating point and complex operations much less
3649 inefficiently. (Thus e.g. you can implement a complex FFT
3651 * The compiler now detects type mismatches between DECLAIM FTYPE
3652 and DEFUN better, and implements CHECK-TYPE more correctly, and
3653 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3654 thanks to patches from Martin Atzmueller.
3655 * various fixes to make the cross-compiler more portable to
3656 ANSI-conforming-but-different cross-compilation hosts (notably
3657 Lispworks for Windows, following bug reports from Arthur Lemmens)
3658 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3659 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3660 fixed thanks to Pierre Mai's CMU CL patches.
3661 * fixes to keep the system from overflowing internal counters when
3662 it tries to use i/o buffers larger than 16M bytes
3663 * fixed bug 45a: Various internal functions required to support
3664 complex special functions have been merged from CMU CL sources.
3665 (When I was first setting up SBCL, I misunderstood a compile-time
3666 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3667 * improved support for type intersection and union, fixing bug 12
3668 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3669 more obscure bugs as well
3670 * some steps toward byte-compiling non-performance-critical
3671 parts of the system, courtesy of patches from Martin Atzmueller
3672 * Christophe Rhodes has made some debian packages of sbcl at
3673 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3674 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3675 stable, but are nonetheless usable. When he's ready, I'd be happy
3676 to add them to the SourceForge "File Releases" section. (And if
3677 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3678 * new fasl file format version number (because of changes in
3679 internal representation of (OR ..) types to accommodate the new
3680 support for (AND ..) types, among other things)
3682 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3683 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3685 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3687 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3688 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3689 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3690 * DESCRIBE now works on structure objects again.
3691 * Most function call argument type mismatches are now handled as
3692 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3693 know whether the function will be redefined before the call is
3694 executed. (The compiler could flag local calls with full WARNINGs,
3695 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3696 it doesn't keep track of enough information to know whether calls
3697 are local in this sense.)
3698 * Compiler output is now more verbose, with messages truncated
3699 later than before. (There should be some supported way for users
3700 to override the default verbosity, but I haven't decided how to
3701 provide it yet, so this behavior is still controlled by the internal
3702 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3703 src/compiler/ir1util.lisp.)
3704 * Fasl file format version numbers have increased again, because
3705 support for the Gray streams extension changes the layout of the
3706 system's STREAM objects.
3707 * The Gray subclassable streams extension now works, thanks to a
3708 patch from Martin Atzmueller.
3709 * The full LOAD-FOREIGN extension (not just the primitive
3710 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3711 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3712 but like most other programs, it defaults to copying the Unix
3713 environment from the original process instead of starting the
3714 new process in an empty environment.
3715 * Extensions which manipulate the Unix environment now support
3716 an :ENVIRONMENT keyword option which doesn't smash case or
3717 do other bad things. The CMU-CL-style :ENV option is retained
3718 for porting convenience.
3719 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3720 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3722 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3724 * A patch from Martin Atzmueller seems to have solved the SIGINT
3725 problem, and as far as we know, signal-handling now works cleanly.
3726 (If you find any new bugs, please report them!)
3727 * The system no longer defaults Lisp source file names to types
3728 ".l", ".cl", or ".lsp", but only to ".lisp".
3729 * The compiler no longer uses special default file extensions for
3730 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3731 seems to expect a single default extension for all compiled code,
3732 and there's no compelling reason to try to stretch the standard
3733 to allow two different extensions.) Instead, byte-compiled files
3734 default to the same extension as native-compiled files.
3735 * Fasl file format version numbers have increased again, because
3736 a rearrangement of internal implementation packages made some
3737 dumped symbols in old fasl files unreadable in new cores.
3738 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3739 many fewer weird special cases.
3740 * Bug #17 (differing COMPILE-FILE behavior between logical and
3741 physical pathnames) has been fixed, and some related misbehavior too,
3742 thanks to a patch from Martin Atzmueller.
3743 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3744 by Tim Moore, ported to SBCL by Martin Atzmueller.
3745 * Martin Atzmueller fixed several filesystem-related problems,
3746 including bug #36, in part by porting CMU CL patches, which were
3747 written in part by Paul Werkowski.
3748 * More compiler warnings in src/runtime/ are gone, thanks to
3749 more patches from Martin Atzmueller.
3750 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3753 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3755 * DESCRIBE now works on CONDITION objects.
3756 * The debugger now handles errors which arise when trying to print
3757 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3759 * The build system now uses an additional file, customize-target-features.lisp,
3760 to allow local modifications to the target *FEATURES* list. (The point of
3761 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3762 debugging features enabled, without having to worry about propagating it
3763 into everyone's system when I do a "cvs update".) When no
3764 customize-target-features.lisp file exists, the target *FEATURES* list
3765 should be constructed the same way as before.
3766 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3767 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3768 warn about behavior which is undefined under the ANSI spec. Note: This
3769 is specified by ANSI, but it's not very popular with programmers.
3770 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3771 macro in the SBCL sources for an example of a workaround which you
3772 might use to make portable ANSI-standard code which does what you want.
3773 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3774 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3775 when it's used as a non-toplevel form. (This is required in order
3776 to implement the DEFCONSTANT-EQX macro.)
3777 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3778 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3779 funny things, due to bugs in the implementation of EVAL-WHEN
3780 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3781 non-toplevel DEFCONSTANTs are uncommon.)
3782 * The core file version number and fasl file version number have been
3783 incremented, because the old noncompliant DEFCONSTANT behavior involved
3784 calling functions which no longer exist, and because I also took the
3785 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3786 * fixed bug 1 (error handling before read-eval-print loop starts), and
3787 redid debugger restarts and related debugger commands somewhat while
3789 ** The QUIT debugger command is gone, since it did something
3790 rather different than the SB-EXT:QUIT command, and since it never
3791 worked properly outside the main toplevel read/eval/print loop.
3792 Invoking the new TOPLEVEL restart provides the same functionality.
3793 ** The GO debugger command is also gone, since you can just invoke
3794 the CONTINUE restart directly instead.
3795 ** The TOP debugger command is also gone, since it's redundant with the
3796 FRAME 0 command, and since it interfered with abbreviations for the
3798 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3800 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3801 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3802 some porting work by Martin Atzmueller)
3803 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3804 with a boatload of patches which he ported to SBCL. Now that those
3806 ** The system tries to make sure that its low-priority messages
3807 are prefixed by semicolons, to help people who like to use
3808 syntax highlighting in their ILISP buffer. (This patch
3809 was originally due to Raymond Toy.)
3810 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3811 patches originally written by Raymond Toy.
3812 ** The compiler understands coercion between single-value and
3813 multiple-VALUES type expressions better, getting rid of some very
3814 weird behavior, thanks to patches originally by Robert MacLachlan
3815 and Douglas Crosher.
3816 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3817 lambda lists, thanks to a patch originally by Pierre Mai.
3818 ** The system no longer bogusly warns about "abbreviated type
3820 ** The compiler gets less confused by inlining and RETURN-FROM,
3821 thanks to some patches originally by Tim Moore.
3822 ** The system no longer hangs when dumping circular lists to fasl
3823 files, thanks to a patch originally from Douglas Crosher.
3824 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3825 undefined function error.
3826 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3827 in structure slot initforms) for now, documented workaround instead:-|
3828 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3829 * fixed bug 5: added stubs for various Gray stream functions called
3830 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3831 installed, at least appropriate type errors are generated
3832 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3833 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3834 consistently in DEFMETHOD forms.
3835 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3836 it doesn't seem to affect SBCL after all
3837 * The C runtime system now builds with better optimization and many
3838 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3840 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3842 * The system is now under CVS at SourceForge (instead of the
3843 CVS repository on my home machine).
3844 * The new signal handling code has been tweaked to treat register
3845 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3846 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3847 downstream, e.g. in debug-int.lisp, has implicit dependencies
3848 on the unsignedness of integer representation of machine words,
3849 and that caused the system to bomb out with infinite regress
3850 when trying to recover from type errors involving signed values,
3851 e.g. (BUTLAST '(1 2 3) -1).)
3852 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3853 (This was one of the bugs Peter Van Eynde reported back in July.)
3854 * The system now uses code inspired by Colin Walters' O(N)
3855 implementation of MAP (from the cmucl-imp@cons.org mailing
3856 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3857 inline the MAP operation, and there is more than one
3858 sequence argument to the MAP call (so that it can't just
3859 do ETYPECASE once and for all based on the type of the
3860 single sequence argument). (The old non-inline implementation
3861 of the general M-argument sequence-of-length-N case required
3862 O(M*N*N) time when any of the sequence arguments were LISTs.)
3863 * The QUIT :UNIX-CODE keyword argument has been renamed to
3864 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3865 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3867 * Raymond Wiker's patches to port dynamic loading from Linux to
3868 FreeBSD have been added.
3869 * The BUGS file is now more nearly up to date, thanks in large part
3870 to Martin Atzmueller's review of it.
3871 * The debugger now flushes standard output streams before it begins
3872 its output ("debugger invoked" and so forth).
3873 * The core version number and fasl file version number have both
3874 been incremented, because of incompatible changes in the layout
3876 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3877 used to not be called for a saved Lisp image.)
3878 * Martin Atzmueller's version of a patch to fix a compiler crash,
3879 as posted on sbcl-devel 13 September 2000, has been installed.
3880 * Instead of installing Martin Atzmueller's patch for the
3881 compiler transform for SUBSEQ, I deleted the compiler transform,
3882 and transforms for some similar consing operations.
3883 * A bug in signal handling which kept TRACE from working on OpenBSD
3885 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3886 away by constant folding
3887 * The system now defines its address space constants in one place
3888 (in the Lisp sources), and propagates them automatically elsewhere
3889 (through GENESIS and the sbcl.h file). Therefore, patching the
3890 address map is less unnecessarily tedious and error-prone. The
3891 Lisp names of address space constants have also been systematized.
3892 * CVS tags like dollar-Header-dollar have been removed from
3893 the sources, because they have never saved me trouble and
3894 they've been source of trouble working with patches and other
3895 diff-related operations.
3896 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3897 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3899 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3901 * The system has been ported to OpenBSD.
3902 * The system now compiles with a simple "sh make.sh" on the systems
3903 that it's supported on. I.e., now you no longer need to tweak
3904 text in the target-features.lisp-expr and symlinks in src/runtime/
3905 by hand, the make.sh takes care of it for you.
3906 * The system is no longer so grossly inefficient when compiling code
3907 involving vectors implemented as general (not simple) vectors (VECTOR T),
3908 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3909 which dares to use the various sequence functions on non-simple
3910 vectors, takes less of a performance hit.
3911 * There is now a primitive type predicate VECTOR-T-P
3912 to test for the (VECTOR T) type, so that e.g.
3913 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3914 can now be compiled with some semblance of efficiency. (The old code
3915 turned the type declaration into a full call to %TYPEP at runtime!)
3916 * AREF on (VECTOR T) is still not fast, since it's still compiled
3917 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3918 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3919 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3920 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3921 without making full calls to SUBTYPEP at runtime.
3922 (Some analogous efficiency issues for non-simple vectors specialized to
3923 element types other than T, or for non-simple multidimensional arrays,
3924 have not been addressed. They could almost certainly be handled the
3925 same way if anyone is motivated to do so.)
3926 * The changes in array handling break binary compatibility, so
3927 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3928 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3929 instead of (VALUES T T).
3930 * By following the instructions that Dan Barlow posted to sbcl-devel
3931 on 2 July 2000, I was able to enable primitive dynamic object
3932 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3933 functionality is not implemented (since it calls ld to resolve
3934 library references automatically, requiring RUN-PROGRAM for its
3935 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3936 try to resolve library references) is now supported.
3937 * The system now flushes the standard output streams when it terminates,
3938 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3939 them at several other probably-convenient times, e.g. in each pass of
3940 the toplevel read-eval-print loop, and after evaluating a form given
3941 as an "--eval" command-line option. (These changes were motivated by a
3942 discussion of stream flushing issues on cmucl-imp in August 2000.)
3943 * The source transform for TYPEP of array types no longer assumes
3944 that an array whose element type is a not-yet-defined type
3945 is implemented as an array of T, but instead punts, so that the
3946 type will be interpreted at runtime.
3947 * There is now some support for cross-compiling in make.sh: each of
3948 the phases of make.sh has its own script. (This should be transparent
3949 to people doing ordinary, non-cross-compile builds.)
3950 * Since my laptop doesn't have hundreds of megabytes of memory like
3951 my desktop machine, I became more motivated to do some items on
3952 my to-do list in order to reduce the size of the system a little:
3953 ** Arrange for various needed-only-at-cold-init things to be
3954 uninterned after cold init. To support this, those things have
3955 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3956 symbols with such names are now uninterned after cold init).
3957 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3958 fasl files for cold load.
3959 ** Remove the old compiler structure pooling code (which used to
3960 be conditional on the target feature :SB-ALLOC) completely.
3961 ** Redo the representation of some data in cold init to be more compact.
3962 (I also looked into supporting byte compiled code at bootstrap time,
3963 which would probably reduce the size of the system a lot, but that
3964 looked too complicated, so I punted for now.)
3965 * The maximum signal nesting depth in the src/runtime/ support code has
3966 been reduced from 4096 to 256. (I don't know any reason for the very
3967 large old value. If the new smaller value turns out to break something,
3968 I'll probably just bump it back up.)
3969 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3971 * Many, many bugs reported by Peter Van Eynde have been added to
3972 the BUGS list; some have even been fixed.
3973 * While enabling dynamic object file loading, I tried to make the
3974 code easier to understand, renaming various functions and variables
3975 with less ambiguous names, and changing some function calling
3976 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3977 * While trying to figure out how to do the OpenBSD port, I tried to
3978 clean up some of the code in src/runtime/. In particular, I dropped
3979 support for non-POSIX signal handling, added various comments,
3980 tweaked the code to reduce the number of compilation warnings, and
3981 renamed some files to increase consistency.
3982 * To support the new automatic configuration functionality in make.sh,
3983 the source file target-features.lisp-expr has been replaced with the
3984 source file base-target-features.lisp-expr and the machine-generated
3985 file local-target-features.lisp-expr.
3986 * fixed a stupid quoting error in make.sh so that using CMU CL
3987 "lisp -batch" as cross-compilation host works again
3989 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3991 * DESCRIBE no longer tries to call itself recursively to describe
3992 bound/fbound values, so that it no longer fails on symbols which are
3993 bound to themselves (like keywords, T, and NIL).
3994 * DESCRIBE now works on generic functions.
3995 * The printer now prints less-screwed-up representations of closures
3996 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3997 * A private symbol is used instead of the :EMPTY keyword previously
3998 used to mark empty slots in hash tables. Thus
3999 (DEFVAR *HT* (MAKE-HASH-TABLE))
4000 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4001 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4002 now does what ANSI says that it should. (You can still get
4003 similar noncompliant behavior if bang on the hash table
4004 implementation with all the symbols you get back from
4005 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4006 This breaks binary compatibility, since tests for equality to
4007 :EMPTY are wired into things like the macroexpansion of
4008 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4010 * There's now a minimal placeholder implementation for CL:STEP,
4011 as required by ANSI.
4012 * An obscure bug in the interaction of the normal compiler, the byte
4013 compiler, inlining, and structure predicates has been patched
4014 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4015 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4017 * Missing ordinary arguments in a macro call are now detected even
4018 when the macro lambda list contains &KEY or &REST.
4019 * The debugger no longer complains about encountering the top of the
4020 stack when you type "FRAME 0" to explicitly instruct it to go to
4021 the top of the stack. And it now prints the frame you request even
4022 if it's the current frame (instead of saying "You are here.").
4023 * As specified by ANSI, the system now always prints keywords
4024 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4026 * The default initial SIZE of HASH-TABLEs is now smaller.
4027 * Type information from CLOS class dispatch is now propagated
4028 into DEFMETHOD bodies, so that e.g.
4029 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4031 is now basically equivalent to
4032 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4033 (DECLARE (TYPE SINGLE-FLOAT X))
4035 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4036 operation, without having to do run-time type dispatch.
4037 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4038 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4039 * Since I don't seem to be making big file reorganizations very often
4040 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4041 is overflowing my ability to conveniently back them up), I've finally
4042 checked the system into CVS. (The CVS repository is on my home system,
4043 not at SourceForge -- putting it on SourceForge might come later.)
4044 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4045 high-level GC-NOTIFY-FOO functions send their output. (There's
4046 still very little control of where low-level verbose GC functions
4047 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4048 less than it used to -- the GC-NOTIFY-FOO functions are now under
4049 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4050 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4051 in only one place in the source code, and propagates it automatically
4052 everywhere that it's needed. Thus e.g. when I bump the version from
4053 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4054 * The C source files now include boilerplate legalese and documentation
4055 at the head of each file (just as the Lisp source files already did).
4056 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4057 to his page will be replaced with a link to his new CLiki service.
4059 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4061 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4062 * The build process now looks for GNU make under the default name "gmake",
4063 instead of "make" as it used to. If GNU make is not available as "gmake"
4064 on your system, you can change this default behavior by setting the
4065 GNUMAKE environment variable.
4066 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4067 can build without error under CMU CL.
4069 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4071 * There is now a partial SBCL user manual (with some new text and some
4072 text cribbed from the CMU CL manual).
4073 * The beginnings of a profiler have been added (starting with the
4074 CMU CL profiler and simplifying and cleaning up). Eventually the
4075 main interface should be through the TRACE macro, but for now,
4076 it's still accessed through vaguely CMU-CL-style functions and macros
4077 exported from the package SB-PROFILE.
4078 * Some problems left over from porting CMU CL to the new
4079 cross-compilation bootstrap process have been cleaned up:
4080 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4081 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4082 part when getting rid of the compiler *BACKEND* stuff.)
4083 ** The value of *NULL-TYPE* was screwed up, because it was
4084 being initialized before the type system knew the final
4085 definition of the 'NULL type. This screwed up several key
4086 optimizations in the compiler, causing inefficiency in all sorts
4087 of places. (I found it because I wanted to understand why
4088 GET-INTERNAL-RUN-TIME was consing.)
4089 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4090 PROCLAIM FTYPE information. Unfortunately this broke binary
4091 compatibility again, since now the forms output by DEFGENERIC
4092 to refer to functions which didn't exist in 0.6.3.
4093 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4094 can use the new (as of 0.6.3) transform for SOME into MAP into
4096 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4097 values to (INTEGER 0 1000000), so that the time code will no longer
4098 occasionally get blown up by Linux returning 1000000 microseconds
4099 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4100 its output conform to the ANSI spec. (Alas, this makes its output
4101 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4102 * A full call to MAP NIL with a single sequence argument no longer conses.
4103 * fixes to problems pointed out by Martin Atzmueller:
4104 * The manual page no longer talks about multiprocessing as though
4105 it were currently supported.
4106 * The ILISP support patches have been removed from the distribution,
4107 because as of version 5.10.1, ILISP now supports SBCL without us
4108 having to maintain patches.
4109 * added a modified version of Raymond Toy's recent CMU CL patch for
4110 EQUALP comparison of HASH-TABLE
4112 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4114 * The system still can't cross-compile itself with
4115 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4116 entails), but at least it can compile more of itself that way
4117 than it used to be able to, and various buglets which were uncovered
4118 by trying to cross-compile itself that way have now been fixed.
4119 * This release breaks binary compatibility again. This time
4120 at least I've incremented the FASL file format version to 2, so that the
4121 problem can be detected reliably instead of just causing weird errors.
4122 * various new style warnings:
4123 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4124 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4125 ** using the lexical binding of a variable named in the *FOO* style
4126 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4127 as specified by ANSI.
4128 * *RANDOM-STATE* is no longer automatically initialized from
4129 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4130 default behavior of the system is to repeat its behavior every time
4131 it's run. If you'd like to change this behavior, you can always
4132 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4133 old convention there was no comparably easy way to get the system to
4134 repeat its behavior every time it was run.
4135 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4136 FTYPE declarations has been removed, in favor of their ANSI
4137 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4138 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4139 the types of their sequence arguments can be determined at compile time.
4140 This is done through a new open code expansion for MAP which eliminates
4141 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4142 argument types can be determined at compile time.
4143 * The optimizer now transforms COERCE into an identity operation when it
4144 can prove that the coerced object is already of the correct type. (This
4145 can be a win for machine generated code, including the output of other
4146 optimization transforms, such as the MAP transform above.)
4147 * Credit information has been moved from source file headers into CREDITS.
4148 * Source file headers have been made more standard.
4149 * The CASE macro now compiles without complaining even when it has
4152 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4154 * (Note that the way that the PCL macroexpansions were rewritten
4155 to accommodate the change in DEFGENERIC below breaks binary
4156 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4157 not run under sbcl-0.6.2. Once we get out of alpha releases,
4158 i.e. hit release 1.0.0, we'll probably try to maintain binary
4159 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4160 and sbcl-1.4.4. Until then, however, it might be fairly common
4161 for maintenance releases to break binary compatibility.)
4162 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4164 * The reporting of unhandled signals has been changed to print some
4165 explanatory text as well as the report form. (Previously only
4166 the report form was printed.)
4167 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4168 it defines, so that the compiler no longer issues undefined function
4169 warnings for compiled-but-not-yet-loaded generic functions.
4170 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4171 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4172 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4173 section of the ANSI spec are supported.
4174 * The "" nickname for the "KEYWORD" package has been removed.
4175 The reader still handles symbol tokens which begin with a package marker
4176 as keywords, but it doesn't expose its mechanism for doing so in the
4177 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4178 * The system now issues STYLE-WARNINGs for contradictory TYPE
4179 proclamations. (Warnings for contradictory FTYPE proclamations would
4180 be nice too, but those can't be done usefully unless the type system
4181 is made smarter about FUNCTION types.)
4182 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4183 systematized, so that "*target-*.lisp is supposed to exist only on the
4184 target and imply that there's a related file which exists on the
4185 host, and *host-*.lisp is supposed to exist only on the host and imply
4186 that there's a related file which exists on the target. This involves a
4187 lot of renaming. Hopefully the acute confusion caused by the renaming
4188 will be justified by the reduction in chronic confusion..
4189 ** runtime-type.lisp -> early-target-type.lisp
4190 ** target-type.lisp -> late-target-type.lisp
4191 ** early-host-format.lisp -> early-format.lisp
4192 ** late-host-format.lisp -> late-format.lisp
4193 ** host-error.lisp -> misc-error.lisp
4194 ** early-error.lisp -> early-target-error.lisp
4195 ** late-error.lisp -> late-target-error.lisp
4196 ** host-defboot.lisp -> early-defboot.lisp
4197 ** code/misc.lisp -> code/target-misc.lisp
4198 ** code/host-misc.lisp -> code/misc.lisp
4199 ** code/numbers.lisp -> code/target-numbers.lisp
4200 ** code/early-numbers.lisp -> numbers.lisp
4201 ** early-host-type.lisp -> early-type.lisp
4202 ** late-host-type.lisp -> late-type.lisp
4203 ** host-typep.lisp -> typep.lisp
4204 ** load.lisp -> target-load.lisp
4205 ** host-load.lisp -> load.lisp
4206 ** host-disassem.lisp -> disassem.lisp
4207 ** host-insts.lisp -> insts.lisp
4208 ** byte-comp.lisp -> target-byte-comp.lisp
4209 ** host-byte-comp.lisp -> byte-comp.lisp
4210 ** host-signal.lisp -> signal.lisp
4211 ** host-defstruct.lisp -> defstruct.lisp
4212 ** late-target-type.lisp -> deftypes-for-target.lisp
4213 Furthermore, several other previously target-only files foo.lisp (e.g.
4214 hash-table.lisp and random.lisp) have been split into a target-and-host
4215 foo.lisp file and a target-only target-foo.lisp file, with their key type
4216 definitions in the target-and-host part, so that the cross-compiler will
4217 know more about target types.
4218 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4219 gone away. In their place are various *BACKEND-FOO* variables
4220 corresponding to the slots of the old structure.
4221 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4222 into the target SBCL has been fixed.
4223 * The chill.lisp system for loading cold code into a running SBCL
4225 * Support for the CMU CL "scavenger hook" extension has been removed.
4226 (It was undocumented and unused in the CMU CL sources that SBCL was
4227 derived from, and stale in sbcl-0.6.1.)
4228 * Various errors in the cross-compiler type system were detected
4229 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4230 (enabling various consistency checks). Many of them were fixed,
4231 but some hard problems remain, so the compiler is back to
4232 running without *TYPE-SYSTEM-INITIALIZED* for now.
4233 * As part of the cross-compiler type system cleanup, I implemented
4234 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4235 * I have started adding UNCROSS calls throughout the type system
4236 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4237 on cross-compiler input files will be able to go away, and various
4239 * CONSTANTP now returns true for quoted forms (as explicitly required
4242 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4244 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4245 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4246 (MAKE-LIST -1). In the longer term, it ought to become true
4247 that declarations are assertions even at SAFETY 1. For now, it's not
4248 quite true even at SAFETY 3, but it's at least more nearly true..
4249 (Note that this change seems to increases the size of the system by
4250 O(5%) and to decrease the speed of the compiler by 20% or more.)
4251 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4252 for the problem of printing dozens of lines of distracting information
4253 about low-level system machinery as part of the top stack frame
4254 on entry to the debugger when an undefined function was called.
4255 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4256 is not reset to COMMON-LISP-USER.
4257 * Compilation of stuff related to dyncount.lisp has been made conditional
4258 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4259 smaller. The various dyncount-related symbols have been moved into
4260 a new "SB-DYNCOUNT" package.
4261 * tty-inspect.lisp has been renamed to inspect.lisp.
4262 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4263 feature has gone away. (When we eventually port to other flavors of
4264 libc and/or Unix, we'll try to make the differences between flavors
4265 invisible at the user level.)
4266 * Various other *FEATURES* tags, and/or their associated conditionals,
4267 have been removed if obsolescent, or given better documentation, or
4268 sometimes given more-mnemonic names.
4270 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4272 * tidied up "make.sh" script
4273 * tidied up system directory structure
4274 * better "clean.sh" behavior
4275 * added doc/FOR-CMUCL-DEVELOPERS
4276 * many many small tweaks to output format, e.g. removing possibly-confusing
4277 trailing #\. character in DESCRIBE-INSTANCE
4278 * (EQUALP #\A 'A) no longer signals an error.
4279 * new hashing code, including EQUALP hashing
4280 * tidied up Lisp initialization and toplevel
4281 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4282 * command line argument processing
4283 * added POSIX-GETENV function to deal with Unix-ish environment variables
4284 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4285 terminating SBCL on EOF
4286 * non-verbose GC by default
4287 * There is no more "sbcl" shell script; the sbcl file is now the C
4288 runtime executable (just like CMU CL).
4289 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4291 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4292 symbol and package information as Lisp native symbol and package objects
4293 instead of strings naming symbols and strings naming packages. This way,
4294 whenever packages are renamed (as in warm init), debug information is
4295 transformed along with everything else.
4296 * tweaked the optimization policy declarations which control the building
4297 of SBCL itself. Now, among other things, the system no longer saves
4298 source location debugging information. (This helps two problems at once
4299 by reducing SBCL size and by keeping SBCL from trying to look for its
4300 sources -- which may not exist -- when reporting errors.)
4301 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4302 debugging and testing purposes
4303 * cleaned up printing, making the printer call PRINT-OBJECT for
4304 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4305 methods, giving nearly-ANSI behavior
4306 * converted almost all special variables to use *FOO* naming convention
4307 * deleted PARSE-TIME functionality, since it can be done portably
4308 * moved some files out of cold init into warm init
4309 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4311 * regularized formatting of source files
4312 * added an install.sh script
4313 * fixed ridiculous memory usage of cross-compiler by making
4314 compiler/alloc.lisp not try to do pooling unless it can hook
4315 itself into the GC of the cross-compilation host. Now the system
4316 builds nicely on my old laptop.
4317 * added :SB-ALLOC in target-features.lisp-expr
4318 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4319 was not implemented)
4320 * re-did condition handling and note reporting in the compiler. Notes
4321 are no longer handled by signalling conditions. Style warnings
4322 and warnings are handled more correctly and reported in such a way
4323 that it's easy to find one or the other in your output (so that you
4324 can e.g. figure out which of many problems caused COMPILE-FILE to
4326 * changed the severity of several compiler warnings from full WARNING
4327 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4328 compiler note reporting so that it doesn't use the condition system
4329 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4330 COMPILE-FILE command)
4331 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4332 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4333 no longer supported, since their implementation was deeply intertwingled
4334 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4335 * removed bogus "support" for compiler macros named (SETF FOO), and
4336 removed the compiler macro for SETF INFO (but only after making a fool
4337 of myself on the cmucl-imp mailing list by posting a bogus patch for
4338 DEFINE-COMPILER-MACRO..)
4339 * Compiled files containing forms which have side effects on the Lisp
4340 reader (such as DEFPACKAGE forms) are now handled more correctly.
4341 (Compiler queuing of top level lambdas has been suppressed by setting
4342 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4343 * deleted various currently-unused source files, e.g. gengc.lisp. They
4344 may be added back at some point e.g. when porting to other architectures,
4345 but until they are it's distracting to distribute them and to try to
4347 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4348 there were so many of them they're just distractions, and UNCROSS is
4349 known to be able to handle the current sources
4350 * moved PROFILE functionality into TRACE, so that it will be clear
4351 how the wrapping and unwrapping of functions when you profile them
4352 interacts with the wrapping and unwrapping of functions when you
4353 trace them. (Actually, the functionality isn't there yet, but at least
4354 the interface specification is there. Hopefully, the functionality will
4355 arrive with some future maintenance release.)
4356 * removed host-oops.lisp
4357 * changed signature of QUIT function to allow UNIX-CODE argument
4358 * fixed READ-SEQUENCE bug
4359 * tweaked verbose GC output so that it looks more like the progress
4360 output that ANSI specifies for functions like LOAD
4361 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4362 * added <http://sbcl.sourceforge.com> to the banner information printed by