1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
3 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
4 atomic compare-and-swap operations on threaded platforms.
5 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
6 allows assining a global minimum value to optimization qualities
7 (overriding proclamations and declarations).
8 * enhancement: closed over variables can be stack-allocated on x86
10 * performance bug fix: GETHASH and (SETF GETHASH) are once again
12 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
13 * bug fix: using obsoleted structure instances with TYPEP and
14 generic functions now signals a sensible error.
15 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
16 (reported by Kristoffer Kvello)
17 * bug fix: backtrace construction is now more careful when making
18 lisp-objects from pointers on the stack, to avoid creating bogus
19 objects that can be seen by the GC.
20 * bug fix: defaulting of values in contexts expecting more than 7
21 variables now works on x86-64. (reported by Christopher Laux)
22 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
24 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
25 as the property-list of a symbol.
26 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
27 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
28 situations requested, are once again file-compileable. (reported
31 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
32 * MOP improvement: support for user-defined subclasses of
33 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
34 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
35 the expansion of DEFMETHOD, is responsible for generating a form
36 which creates a list of specializers when evaluated. Additional
37 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
38 debugging and introspective support.
39 * minor incompatible change: the (unsupported) spinlock interface
40 has changed: free spinlock now has the value NIL, and a held spinlock
41 has the owning thread as its value.
42 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
43 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
45 * enhancement: name of a socket-stream is now "a socket" instead of
47 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
48 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
49 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
50 for generic functions. (Thanks to Tobias C. Rittweiler)
51 * optimization: bignum printing speed has been improved by 20-40%
52 (depending on the bignum size.)
53 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
55 * bug fix: the cache used by the CLOS to store precomputed effective
56 methods, slot offsets, and constant return values is now thread and
58 * bug fix: generic function dispatch function updating is now thread
59 and interrupt safe (in the sense that the known issues have been
61 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
62 * bug fix: interning EQL-specializers is now thread and interrupt safe.
63 * bug fix: asdf systems with dependencies to the SB-POSIX or
64 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
65 * bug fix: interrupt safety of applicable method computation has been
68 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
69 * new contrib: sb-cover, an experimental code coverage tool, is included
71 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
73 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
74 produce more efficient code on x86-64 if the compiler can determine
75 that (+ INDEX <constant>) does not require a bounds check and FOO
76 has an element type at least 8 bits wide.
77 * enhancement: a new, experimental synchronous timeout facility is
78 provided. Refer to SB-SYS:WITH-DEADLINE for details.
79 * enhancement: when a symbol name conflict error arises, the
80 conflicting symbols are always printed with a package prefix.
81 (thanks to Kevin Reid)
82 * enhancement: stepping is now once again supported on the SPARC and
83 MIPS platforms. (It is also now more likely to work on CheneyGC
85 * enhancement: sb-sprof can now also track and report accurate call
87 * bug fixes: the treatment of non-standard subclasses of
88 SB-MOP:SPECIALIZER is more correct.
89 * incompatible change: PURIFY no longer copies the data from the
90 dynamic space into the static and read-only spaces on platforms
91 that use the generational garbage collector
92 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
94 * bug fix: GC race condition occasionally resulting in crashes with
95 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
97 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
99 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
100 a STYLE-WARNING to be signalled (regression from 1.0.4.)
101 * bug fix: an asynchronous interrupt could previously leave the
102 system running with GC inhibited.
103 * bug fix: a DECLARE form evaluated at top-level now causes an error
104 rather than silently (or verbosely) returning NIL.
105 * bug fix: trying to dissassemble functions compiled at high DEBUG could
106 result in a error being signalled due to source form lookup errors.
107 (reported by Peter Graves)
109 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
110 * incompatible change: removed writer methods for host-ent-name,
111 host-ent-addresses -- changing the values did not update the DNS
113 * minor incompatible change: changed experimental JOIN-THREAD interface
114 * documentation: the manual now lists reader and writer methods
115 in class slot documentation sections. (thanks to Richard M Kreuter)
116 * documentation: unwinding from asyncronous events has been
117 documented as unsafe.
118 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
119 in multithreaded application code.
120 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
121 platforms. (thanks to James Anderson for the optimization hint)
122 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
124 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
125 variants no longer cons.
126 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
127 their NOT- variants no longer cons.
128 * optimization: EQUAL hash tables no longer use SXHASH for objects
129 of all data types, but instead use an EQL hash for types for which
130 EQUAL is the same as EQL.
131 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
132 are significantly faster.
133 * optimization: new STRING-OUTPUT-STREAM implementation is much
134 faster for large outputs and conses 30% less on average.
135 * enhancement: XREF information is now collected for references made
136 to global variables using SYMBOL-VALUE and a constant argument.
137 * enhancement: SIGINT now causes a specific condition
138 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
139 * bug fix: FILE-POSITION used to signal an error in some cases where
140 ANSI requires it to return NIL.
141 * bug fix: ADJUST-ARRAY is now interrupt-safe.
142 * bug fix: adding and removing fd-handlers is now interrupt-safe.
143 * bug fix: inlined calls to C now ensure 16byte stack alignment on
145 * bug fix: bad type declaration in the CLOS implementation has
146 been fixed. (reported by James Anderson)
147 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
148 (thanks to Lutz Euler)
149 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
150 required by the CLHS. (reported by Samium Gromoff)
151 * bug fix: dead unbound variable references now signal an error.
152 (reported by Marco Monteiro)
153 * bug fix: / with an unused value was being deleted in safe code.
154 (thanks to Marco Monteiro and Kevin Reid)
155 * bug fix: number of characters that can be written onto a single
156 line in a file is unlimited.
157 * bug fix: some GC deadlocks caused by asynchronous interrupts have
158 been fixed by inhibiting interrupts for when GC is disbled.
159 * bug fix: some interrupt safety issues with GETHASH, (SETF
160 GETHASH), CLRHASH and REMHASH have been fixed.
161 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
162 specifier no longer causes infinite recursion.
163 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
164 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
165 * bug fix: an &environment argument with the correct variable information
166 is passed to macros that are expanded during byte compilation.
167 (reported by Samium Gromoff)
168 * bug fix: a logic error was causing the hash function for bignums to
169 have excessive amounts of collisions. (reported by Faré Rideau)
170 * bug fix: modifying the contents of an array could change the return
171 value of SXHASH on that array, which is only allowed for strings
172 and bit vectors. (bug introduced in 0.9.16)
173 * improvement: the x86-64/darwin port now passes all tests (except
174 for the debugger tests) but should still be considered
175 experimental until this is fixed.
176 * improvement: a style-warning is signaled for CASE (etc) clauses with
177 duplicate keys. (thanks to Kevin Reid)
178 * improvement: macos/x86 and macos/x86-64 now use mach exception
179 handlers for dealing with illegal instructions (for trapping and
180 error handling) and memory protection violations (for GC).
182 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
183 * new platform: experimental support for x86-64/darwin (MacOS).
184 * incompatible change: the thread-safe (on most platforms) getaddrinfo
185 and getnameinfo sockets functions are used instead of gethostbyaddr
186 and gethostbyname, on platforms where the newer functions are available.
187 As a result, the ALIASES field of HOST-ENT will always be NIL on these
189 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
190 don't print the startup banner, but behave as if --noinform was passed
191 as a command line argument. (thanks to Kevin Reid)
192 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
193 * optimization: code using alien values with undeclared types is much faster.
194 * optimization: the compiler is now able to open code SEARCH in more cases.
195 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
196 * bug fix: using standardized COMMON-LISP special variables as loop
197 variables no longer signals bogus package lock violations. (reported
199 * bug fix: declaring local loop variables to be of a range-limited type
200 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
201 (reported by Andras Simon)
202 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
203 bugs remain on x86-64.)
204 * bug fix: the #= and ## reader macros now interact reasonably with
205 funcallable instances.
206 * bug fix: type-checks for function arguments were compiled using the
207 compiler policy settings of the wrong lexical environment (in 1.0.2
209 * bug fix: SHADOW accepts characters as string designators, as required
210 by the spec (thanks to Eric Marsden)
211 * bug fix: fixed GC safety issues when foreign functions are called with
212 non-base strings as arguments
213 * bug fix: more consistent error output for fatal-compiler-warnings, like
215 * bug fix: functions with proper names were showing up as NIL in
216 backtraces (reported by Edi Weitz, regression in 1.0.2)
218 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
219 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
220 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
221 produce more efficient code on the x86 if the compiler can determine
222 that (+ INDEX <constant>) does not require a bounds check and FOO
223 has an element type at least 8 bits wide.
224 * bug fix: references to undefined variables are handled the same way
225 in toplevel forms as in the normal compiler
226 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
228 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
229 (thanks to Jon Buller)
230 * improvement: faster compilation times for complex functions
231 * improvement: added readlink support to SB-POSIX (thanks to Richard
234 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
235 * improvement: experimental support for mach exception handling on
236 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
238 * improvement: support for GBK external format.
239 (thanks to Chun Tian (binghe))
240 * improvement: the debugger now displays variables that have been closed
241 over, in code compiled with (DEBUG 2) or higher
242 * improvement: support for executable cores on NetBSD (thanks to
244 * new feature: added a RESTART-FRAME debugger command
245 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
246 be used to provide an implementation for FILE-POSITION on Gray streams
247 (thanks to Eric Marsden)
248 * new feature: add syslog, openlog and closelog support to SB-POSIX
249 (thanks to Richard Kreuter)
250 * optimization: the function call overhead in code compiled with
251 a high DEBUG optimization setting is significantly
252 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
253 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
254 (thanks to Tony Martinez)
255 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
256 evaluated once (reported by Kevin Reid)
257 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
258 works again (thanks to Kilian Sprotte)
259 * bug fix: an error is signaled for tagbodies with duplicate tags
260 (thanks to Stephen Wilson)
261 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
262 * bug fix: Win32 port can now handle foreign code unwinding Lisp
263 stack frames from alien callbacks.
264 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
265 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
266 65k pages (thanks to David Woodhouse)
267 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
269 changes in sbcl-1.0.1 relative to sbcl-1.0:
270 * new platform: FreeBSD/x86-64, including support for threading.
271 * new feature: the compiler stores cross-referencing information
272 abount function calls (who-calls), macroexpansion (who-macroexpands)
273 and special variables (who-binds, who-sets, who-references) for code
274 compiled with (< SPACE 3). This information is available through the
275 sb-introspect contrib.
276 * new feature: users may subclass SEQUENCE, and have instances of
277 these classes interoperate with standard Common Lisp functions if
278 a number of methods are defined. (This feature is experimental
279 and the interface subject to change based on feedback from SBCL
280 users and the general community)
281 * improvement: sb-sprof traces call stacks to an arbitrary depth on
282 x86/x86-64, rather than the previous fixed depth of 8
283 * improvement: another pthread back-end of mutex "pthread-futex".
284 * bug fix: non-ascii command-line arguments are processed correctly
285 (thanks to Yaroslav Kavenchuk)
286 * bug fix: non-required arguments were not passed correctly when a method
287 defined using DEFMETHOD was called from a mop-generated method using
288 CALL-NEXT-METHOD (reported by Pascal Costanza)
289 * bug fix: recursion is now permitted in accessors through
290 SLOT-UNBOUND. (reported by Pascal Costanza)
291 * bug fix: an error was signaled at startup if the HOME environment
292 variable was defined, but had an empty value (reported by Peter Van Eynde)
293 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
294 signal a PROGRAM-ERROR, not a BUG.
295 * bug fix: SB-POSIX:READDIR works when built with large file support.
296 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
298 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
299 for better automatic indentation support. (thanks to Matt Pillsbury)
300 * optimization: loading generic functions no longer takes O(n^2) time,
301 proportional to the amount of methods in the generic function
302 (reported by Todd Sabin and Jeremy Brown)
303 * optimization: the FIND and POSITION family of sequence functions
304 are significantly faster on arrays whose element types have been
306 * improvements to the Windows port:
307 ** Intermittent heap corruption problems have been fixed. (thanks
308 to Alastair Bridgewater)
309 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
311 ** Lisp is able to unwind foreign exception frames from alien
312 callbacks. (thanks to Alastair Bridgewater)
314 changes in sbcl-1.0 relative to sbcl-0.9.18:
315 * improvement: experimental support for threading on FreeBSD/x86.
316 (thanks to NIIMI Satoshi)
317 * improvement: runtime option --dynamic-space-size can be used
318 to set the size of the dynamic space reserved on startup.
319 * improvement: floating point modes in effect are now saved in
320 core, and restored on startup.
321 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
322 startup, not time since first call to GET-INTERNAL-REAL-TIME.
323 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
324 threads are not running after *SAVE-HOOKS* have run.
325 * improvement: writes to CLOS instance slots are type-checked in code
326 compiled with (SAFETY 3)
327 * improvement: floating-point exception handling on FreeBSD (thanks to
329 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
330 (thanks to Zach Beane)
331 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
333 * improvement: added support for the Shift-JIS external format.
334 (contributed by NIIMI Satoshi)
335 * improvement: callbacks are supported on Linux/PPC. (thanks to
337 * bug fix: compiler bug triggered by a (non-standard) VALUES
338 declaration in a LET* was fixed. (reported by Kaersten Poeck)
339 * bug fix: file compiler no longer confuses validated and already
340 dumped structurres. (reported by Kaersten Poeck)
341 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
342 fill-pointer signals a type-error as required. (thanks to
344 * bug fix: disassemly of funcallable instances works.
345 * bug fix: single stepping on PPC.
346 * bug fix: fix thread-safety problems in the type system (generally
347 manifesting as nonsensical errors like "STRING is a bad type specifier
348 for sequences" or "The value 1 is not of type FIXNUM")
349 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
350 (reported by Josip Gracin)
351 * bug fix: an error is signaled for attempts to displace arrays with
352 incompatible element types (thanks to Mario Mommer)
353 * bug fix: more correct handling of wide characters in debug info
354 (bug reported by Attila Lendvai and fixed by Juho Snellman)
355 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
356 and don't cause extra consing
357 * optimization: MAP and MAP-INTO are significantly faster on vectors
358 whose elements types have been declared.
359 * Improvements to SB-SPROF:
360 ** Support for allocation profiling
361 ** Reduced profiling overhead, especially for long profiling runs
362 * Improvements to the Windows port:
363 ** floating point exceptions are now reported correctly.
364 ** stack exhaustion detection works partially.
365 ** more accurate GET-INTERNAL-REAL-TIME.
366 ** better RUN-PROGRAM behaviour in case of errors in the child
368 ** PROBE-FILE now simplifies pathnames correctly.
369 ** DIRECTORY now works correctly with :WILD-INFERIORS.
370 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
371 1st 1970 (although time timezone data will be unreliable).
372 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
374 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
375 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
376 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
378 * bug fix: two potential GC deadlocks affecting threaded builds.
379 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
380 returns T (reported by Anton Kazennikov)
381 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
382 with non-variable places
383 * bug fix: remove a race condition in the setting of
384 funcallable-instance functions, this should make threaded CLOS
385 code more stable against memory faults.
386 * bug fix: corruption of specials when unbinding is interrupted by an
387 asynchronous unwind (reported by Hannu Koivisto)
388 * improvement: the debugger will now also display local variables that
389 are only used once, for code compiled with a DEBUG optimization quality
392 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
393 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
394 * incompatible change: External-format support for FFI calls. The
395 SB-ALIEN:C-STRING no longer implies an ASCII
396 external-format. Instead the string is subject to external-format
397 conversion. Additionally return values of type SB-ALIEN:C-STRING
398 are by default of type (SIMPLE-ARRAY CHARACTER), not
399 SIMPLE-BASE-STRING. For an alien type that behaves like the old
400 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
401 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
402 * incompatible change: SB-EXT package no longer contains the
403 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
404 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
406 * incompatible change: the single-stepper is no longer available
407 on Alpha, Mips and Sparc platforms.
408 * minor incompatible change: the direct superclasses of
409 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
410 not (STANDARD-OBJECT FUNCTION). This makes the
411 class-precedence-lists of GENERIC-FUNCTION and
412 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
414 * minor incompatible change: the default stream external format on
415 non-unicode SBCL is no longer detected from the locale, but is always
417 * new feature: Add a version of evaluator that uses an interpreter instead
418 of the compiler. EVAL still uses the compiler by default, to switch it
419 to use the interpreter, set the value of the variable
420 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
421 * minor incompatible change: the single-stepper REPL has been merged
422 with the normal debugger (see the "Stepping" heading of the debugger help
423 for more details). The debugger command STEP will no longer switch
424 to the single-stepper REPL.
425 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
426 for a type now works.
427 * bug fix: loading of default sysinit file works. (thanks to Leonid
429 * bug fix: better detection of circularities in the file-compiler.
430 (reported by Marco Monteiro)
431 * bug fix: the CL pathname functions now work with files that have
432 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
433 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
434 systems with Unix98 pty semantics.
435 * bug fix: ASDF-INSTALL will now work with bsd tar.
436 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
438 * bug fix: timers expiring in dead threads no longer cause a
439 type-error (reported by Paul "Nonny Mouse").
440 * bug fix: thanks to more lightweight single-stepper instrumentation,
441 code compiled with (DEBUG 3) will compile and execute significantly faster,
442 and will have more accurate type-inferencing than before
443 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
444 whose bindings are modified
445 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
446 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
447 ** replace dummy implementations of CL:MACHINE-INSTANCE and
448 CL:SOFTWARE-VERSION with proper version
450 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
451 * feature: implemented the READER-METHOD-CLASS and
452 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
453 as specified by AMOP.
454 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
456 * optimization: faster LOGCOUNT implementation on x86 and x86-64
457 (thanks to Lutz Euler)
458 * optimization: hashing of general arrays and vectors has been
459 improved. (reported by Any Fingerhut)
460 * enhancement: SB-INTROSPECT is now able to find definitions of
461 profiled functions. (thanks to Troels Henriksen)
462 * enhancement: compiler-macro expansion applies now to FUNCALL forms
463 as well, allowing compiler-macros for SETF-functions to expand.
464 * enhancement: step-instrumentation no longer wraps known
465 single-value functions with multiple-value context, allowing
466 better type inference.
467 * fixed bug #337: use of MAKE-METHOD in method combination now works
468 even in the presence of user-defined method classes. (reported by
469 Bruno Haible and Pascal Costanza)
470 * fixed bug #339(c): if there are applicable methods not part of any
471 long-form method-combination group, call INVALID-METHOD-ERROR.
472 (reported by Bruno Haible)
473 * fixed bug #361: the :FUNCTION initarg in the protocol for
474 initialization of methods can now be used to override
475 internally-produced optimized functions. (reported by Bruno
477 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
478 system-provided lambda expression no longer cause warnings about
479 unbound #:|pv-table| symbols.
480 * bug fix: improved the handling of type declarations and the
481 detection of violations for keyword arguments with non-constant
483 * bug fix: potentially erronous calls to PATHNAME and
484 MERGE-PATHNAMES were being flushed in some cases.
485 (reported by Richard Kreuter)
486 * bug fix: compiled calls to TYPEP were mishandling obsolete
487 instances. (reported by James Bielman and Attila Lendvai)
488 * bug fix: format strings with ~> without matching ~< no longer
489 trigger an AVER, but signal an understandable error instead.
490 (reported by Antonio Martinez)
491 * bug fix: specifying an output-file without a pathname-type for
492 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
493 of input-file instead of "fasl". (reported by Robert Dodier)
494 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
495 (reported by James Y Knight).
496 * bug fix: compiler-macros-function did not consider the environment
497 argument for shadowing by local functions.
498 * bug fix: compiler-macros expansion was inhibited by local INLINE
500 * bug fix: inline expansions of known functions were subject to
501 step-instrumentation in high debug policies, leading to problems
503 * bug fix: compiler failed to differentiate between different CONS
505 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
506 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
507 * bug fix: #368: incorrect use of expressed vs. upgraded array
509 * bug fix: #306a: more precise unions of array types.
510 * thread-safety improvements:
511 ** CONDITION-WAIT could return early on Linux, if the thread was
512 interrupted and subsequently continued with SIGCONT.
513 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
515 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
516 * added support for the ucs-2 external format. (contributed by Ivan
518 * minor incompatible change: pretty printing of objects of type
519 (cons symbol) is, in the default pprint-dispatch-table, now
520 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
522 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
523 called later in a class's lifetime, possibly as late as when the
524 first instance of the class is created. Previously,
525 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
526 class became finalizeable.
527 * fixed bug: FILE-POSITION sometimes returned inconsistent results
528 for multibyte external-format streams. (thanks to "vbzoli")
529 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
530 with :ALLOCATION :CLASS inherited from superclasses of the
532 * fixed bug: anonymous classes can now be created using the :NAME
533 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
534 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
535 * fixed bug: core-files saved with :EXECUTABLE T can again be
536 executed when SBCL_HOME isn't set. (reported by James Knight)
537 * fixed bug: toplevel LOCALLY forms with declarations could
538 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
539 * fixed bug: printing from several different threads using different
540 values of *print-case* could cause invalid output, due to
541 some internal special variables of the printer not being bound
542 thread-locally (reported by Max Mikhanosha)
543 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
544 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
545 (reported by Pascal Costanza)
546 * fixed bug: FINALIZE-INHERITANCE is called from
547 REINITIALIZE-INSTANCE on classes when the class has previously
548 been finalized, as required by AMOP.
549 * minor code generation optimizations:
550 ** better register allocation in CLOS dispatching functions
551 ** overflow detection when coercing signed bytes to fixnums on x86-64
552 ** is now implemented with one IMUL instruction instead of three shifts
553 ** more efficient bit-vector access on x86 and x86-64
554 ** more efficient access to raw structure slots on x86 and x86-64
555 * fixed some bugs revealed by Paul Dietz' test suite:
556 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
559 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
560 * feature: thread support on Solaris/x86, and experimental thread support
562 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
564 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
565 default method for SLOT-UNBOUND.
566 * minor incompatible change: prevent the user from specializing the
567 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
568 somewhat counter to the intent of the protocol, I (CSR) think, and
569 additionally it just doesn't work in SBCL as currently
570 implemented, thanks to optimizations (that are always valid for
571 the other three SLOT-VALUEish functions, but not for the setter).
572 * minor incompatible change: the :SB-LDB feature is now enabled by
573 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
574 the low-level debugger.
575 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
576 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
577 down to an SB-SYS:FD-STREAM, allowing direct communication in
578 more cases. Temporary files and pipes are still used as a fallback
580 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
581 on the non-reentrant C functions localtime(3) and gmtime(3).
582 * bug fix: global optimization policy was not visible in LOCALLY and
584 * bug fix: class objects can be used as specializers in methods.
585 (reported by Pascal Costanza)
586 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
587 with a type error. (reported by blitz_ on #lisp)
588 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
589 failed with a type error. (reported by Pascal Bourguignon)
590 * bug fix: merging pathnames against defaults with :DIRECTORY
591 starting with '(:RELATIVE :BACK) should preserve the :BACK.
592 (reported by James Y Knight)
593 * bug fix: saving large (>2GB) cores on x86-64 now works
594 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
595 constant index and a (simple-array (signed-byte 32)) array
596 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
597 error (patch by Robert J. Macomber)
598 * feature: TIME now displays the amount of run-time spent in GC
599 * bug fix: The debugger now does a better job of respecting
600 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
601 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
602 workaround for bug 403.)
603 * bug fix: floating point exception handling now works on FreeBSD
604 (thanks to NIIMI Satoshi)
605 * fixed some bugs revealed by Paul Dietz' test suite:
606 ** MISC.641: LET-conversion were not supposed to work in late
608 * improvements to the Win32/x86 port:
609 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
610 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
612 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
613 * new feature: source path information is generated for macro-expansion
614 errors for use in IDE's like Slime (thanks to Helmut Eller)
615 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
616 * bug fix: compiler does not loop forever on an invalid type in
618 * improvement: compilation of most CLOS applications is significantly
620 * optimization: added a limited bytecode compiler for simple toplevel
621 forms, speeding up compilation and FASL loading
622 * bug fix: the statistical profiler now properly distinguishes anonymous
625 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
626 * minor incompatible change: in sbcl-0.9.11 (but not earlier
627 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
628 into subclasses of the finalized class. Now user calls to
629 FINALIZE-INHERITANCE finalize just the one class, and calls by the
630 system return before any subclasses are finalized.
631 * minor incompatible change: The reader no longer ignores errors
632 regarding non-existent packages in #+ and #- feature tests.
633 * new feature: command line options --no-sysinit, --no-userinit to
634 inhibit loading the corresponding init files
635 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
636 for sending data through UDP sockets (thanks to François-René Rideau)
637 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
638 error is signalled instead (thanks to François-René Rideau)
639 * minor incompatible change: Error signalling behaviour of lexical
640 operations violating package locks has changed slightly. Refer to
641 documentation on package locks for details.
642 * enhancement: EVAL can process IF-expressions without resorting to the
644 * bug fix: Some operations involving SIN, COS, or TAN weren't being
645 constant-folded properly due to missing out-of-line functions.
646 (reported by Mika Pihlajamäki)
647 * bug fix: LISTEN sometimes returned T even in cases where no data was
648 immediately available from the stream
649 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
650 were reversed. (reported by Levente Mészáros)
651 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
652 when given a vector displaced to an adjustable array. (reported
654 * bug fix: derivation of float boundaries from numbers outside the
655 appropriate float range (reported by John Wiseman)
656 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
658 * bug fix: precomputing cache entries for generic functions with
659 some subclasses of specializers as yet invalid does not attempt to
660 fill a cache line with a negative offset. (reported by Levente
662 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
663 allow condition class objects as arguments to DOCUMENTATION and
664 (SETF DOCUMENTATION); only find and set documentation for
665 structure names for the STRUCTURE doc-type. (suggested by Gary
667 * improvements to the Win32/x86 port:
668 ** (user-homedir-pathname) and default initialization file
669 locations now know about the user's "Documents and Settings"
670 directory (thanks to Yaroslav Kavenchuk)
671 ** run-program is implemented (thanks to Mike Thomas)
672 ** sockets support (thanks to Timothy Ritchey)
673 ** better backtrace support (thanks to Alastair Bridgewater)
674 ** sb-grovel supported
675 ** asdf-install and sb-posix work somewhat
676 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
677 * improvements to the Solaris/x86 port:
678 ** works on Solaris 11/Solaris Express
679 ** floating-point exception handling support
680 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
681 * fixed some bugs revealed by Paul Dietz' test suite:
682 ** REMOVE-METHOD returns its generic function argument even when
683 the method is not one of the generic functions' methods.
684 ** objects declared as MEMBER types can be admissible arguments to
686 ** printing characters should simply be printed by the FORMAT ~:C
688 ** compiler failure when compiling functions with hairy constant
689 defaults for optional parameters.
690 ** compiler produces wrong code when MAYBE-INLINE-expanding a
691 function, which is already optimized.
693 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
694 * new platform: experimental support for SBCL x86/Darwin, including
695 MacOS X 10.4.5 on Intel.
696 * new feature: Unicode character names are now known to the system
697 (through CHAR-NAME and NAME-CHAR).
698 * new feature: the filesystem location of SBCL's core file is
699 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
700 * minor incompatible change: the contrib modules SB-POSIX and
701 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
702 this change is to make it easier to distribute
703 location-independent binaries.
704 * bug fix: as implied by AMOP, standardized classes no longer have
705 slots named by external symbols of public packages. (reported by
707 * optimization: calling functions via constant symbols -- as in
708 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
709 via the function object as in (FUNCALL #'FOO).
710 * optimization: on x86 and x86-64, the calling convention for the
711 case of dealing with an unknown number of values has been altered
712 to be friendlier to the prediction heuristics implemented,
713 particularly with reference to CALL and RET pairing. (thanks to
714 Alastair Bridgewater)
715 * optimization: on x86, the code for access to thread-local symbol
716 values and binding/unbinding of thread-local symbols is smaller.
717 (thanks to Alastair Bridgewater)
718 * enhancement: CONSTANTP is now able to determine constantness of
719 more complex forms, including calls to constant-foldable standardized
720 functions and some special forms beyond QUOTE.
721 * fixed bug: occasional GC crashes on Solaris/x86
722 * optimization: x86-64 supports stack allocation of results of simple
723 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
724 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
725 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
726 (thanks to James Knight)
727 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
728 by the standard (thanks to Stephen Compall)
730 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
731 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
732 be used for bundling the runtime and the core file into one
733 executable binary. This feature is not currently supported on all SBCL
734 platforms. (thanks to James Bielman and NIIMI Satoshi)
735 * new feature: a generational or ephemeral garbage collector is now
736 the default on the PowerPC platform (both Linux and Darwin). The
737 old Cheney (stop and copy) collector is a build-time option.
738 (thanks to Cyrus Harmon, after Raymond Toy)
739 * minor incompatible change: the method by which SBCL finds its
740 contributed modules has changed; it no longer relies on symbolic
741 links from an $SBCL_HOME/systems directory, but searches directly
742 in the subdirectories of $SBCL_HOME.
743 * enhancement: the dynamic heap size on the Linux/PPC platform is
744 markedly larger, even using the older Cheney garbage collector.
745 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
746 certain complicated string types. (reported by Gary King)
747 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
748 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
749 * fixed bug: DOCUMENTATION on structures no longer signals an
750 error if no structure type of the right name exists. (reported by
752 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
753 applicable non-standard primary or :AROUND method on
754 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
755 no longer cause unbound variable errors. (reported by Kevin Reid)
756 * fixed bug: metacircle resolution in cases where methods have slots
757 added before the slots from STANDARD-METHOD. (reported by Jean
759 * fixed bug: the Power PC assembly code for calling into lisp
760 sometimes computed its return address wrongly (depending on the
761 15th bit of the address, fixed at link time). This used to
762 manifest itself in a segmentation violation while building PCL.
763 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
764 many others over the years)
765 * fixed bug: ROOM no longer reports silly numbers for stack usage if
766 the stack pointer (treated as a signed integer) is negative.
767 (thanks to Peter van Eynde)
769 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
770 * new platform: experimental support for the Windows operating
771 system has been added. (thanks to Alastair Bridgewater)
772 * fixed several bugs in and robustified the PPC FFI (including
773 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
774 * bug fix: allow non-simple string symbol names (reported by Paul
776 * bug fix: interrupt handling on NetBSD (thanks to Richard M
778 * bug fix: saving a core corrupted callbacks on x86/x86-64
779 * bug fix: closed a loophole in metacircularity detection and
780 grounding in the PCL implementation of CLOS.
781 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
782 (reported by tomppa on #lisp)
783 * bug fix: clear the direction flag on Lisp -> C transitions, as
784 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
785 (reported by Andrej Grozin and Hendrik Maryns)
786 * optimization: major improvements to GC efficiency on GENCGC platforms
787 * optimization: faster implementation of EQUAL
788 * optimization: emit more efficient opcodes for some common
789 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
790 * optimization: several other minor code-generation improvements on x86-64
791 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
793 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
794 * minor incompatible change: (SETF CLASS-NAME) and (SETF
795 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
796 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
797 by ANSI as a generic function, and for consistency (SETF
798 GENERIC-FUNCTION-NAME) is treated likewise.
799 * fixed bug #233.b: make constraint propagation notice when a variable
800 value is changed after it is referenced but before it is used
801 * fixed bug #296: no more arbitrary behaviour from filenames with
802 odd characters as --load arguments. (reported by Adam Warner)
803 * fixed bug #390: :CHARACTER-SET pathname components now work as
804 expected. (reported by Tim Daly Jr)
805 * fixed bug #391: complicated :TYPE intersections in slot
806 definitions no longer cause an error in PCL internals.
807 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
808 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
809 returning the number of octets which would be written to the
810 file-stream. (thanks to Robert J. Macomber)
811 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
812 arguments to the internal stream functions. (thanks to David
814 * bug fix: stack exhaustion is now handled on the x86/SunOS
815 platform. (thanks to Daisuke Homma)
816 * bug fix: more accurate ROOM results on GENCGC platforms
817 * optimization: improved type inference for arithmetic-for
818 index variables in LOOP
819 * optimization: faster floating-point SQRT on x86-64
820 * fixed some bugs revealed by Paul Dietz' test suite:
821 ** DOCUMENTATION returns NIL instead of "" for method combinations
822 that don't have a docstring
824 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
825 * minor incompatible change: (SETF CLASS-NAME) and (SETF
826 GENERIC-FUNCTION-NAME) are no longer generic functions, and
827 therefore are not customizeable by user code (as seems to be at
828 least permitted and maybe required by AMOP). As a consolation,
829 however, the SBCL implementation of these functions now calls
830 REINITIALIZE-INSTANCE as specified by AMOP.
831 * bug fix: slot-definition documentation is propagated as per ANSI
832 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
833 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
834 argument, as do eslotd initialization methods. (from Pascal
835 Costanza's "Closer" project)
836 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
837 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
839 * bug fix: it is now possible to have more than one subclass of
840 STANDARD-GENERIC-FUNCTION without causing stack overflow.
841 (reported by Bruno Haible, Pascal Costanza and others)
842 * bug fix: the dependent update protocol now works for generic
843 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
845 * bug fix: condition-class instances corresponding to
846 DEFINE-CONDITION forms are now created eagerly. (reported by
847 Kalle Olavi Niemitalo on comp.lang.lisp)
848 * bug fix: floating point printing is more accurate in some
849 circumstances. (thanks to Simon Alexander)
850 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
851 merged with *DEFAULT-PATHNAME-DEFAULTS*.
852 * bug fix: callbacks on OS X now preserve stack-alignment.
853 (thanks to Cyrus Harmon)
854 * enhancement: the x86-64 disassembler is much better at
855 disassembling SSE instructions. (thanks to Lutz Euler)
856 * enhancement: CHECK-TYPE on a variable now allows type-inference
857 (thanks to Vincent Arkesteijn)
858 * optimization: improved performance of EUC-JP external format.
859 (thanks to NIIMI Satoshi)
860 * optimization: performance improvements to IO on file streams of
861 :ELEMENT-TYPE CHARACTER
862 * optimization: much faster memory allocation on x86-64
863 * optimization: faster garbage collections (latency and throughput)
865 * optimization: faster bignum subtraction and division on x86 and x86-64
866 (thanks to Lutz Euler)
867 * optimization: more accurate type-inference for loops that use a
868 floating point index variable or a negative step.
870 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
871 * bug fix: add a workaround to SBCL looping infinitely at startup on
872 Linux kernels with apparently buggy implementations of personality().
873 (thanks to Svein Ove Aas)
874 * bug fix: Unicode symbols are correctly printed in LDB backtraces
875 (thanks to David Lichteblau)
876 * bug fix: local bindings shadow global symbol macros for
877 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
878 * optimization: non-open coded uses of numeric comparison operators
879 (e.g. >) no longer cons when called with more than one parameter
880 on platforms supporting dynamic-extent allocation.
881 * enhancement: saving cores with foreign code loaded is now supported
882 on MIPS/Linux in addition to the previously supported platforms.
883 * bug fix: FIND-RESTART now tests for activity, not applicability when given
884 a restart object as identifier. (reported by Helmut Eller for CMUCL)
885 * bug fix: division by zero in sb-sprof when no samples were collected
886 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
887 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
888 (thanks to Alexander Kjeldaas)
889 * new feature: ignoring the return values of destructive functions
890 like NREVERSE signals a compile-time style-warning.
891 (thanks to Kevin Reid)
892 * new feature: the alignment of alien structure fields can be
893 explicitly specified. (thanks to Cyrus Harmon)
894 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
895 and dump core on SIGQUIT
897 ** incompatible change: threads do not inherit values of specials
898 from their parents (see manual)
899 ** bug fix: threads stacks belonging to dead threads are freed by the
900 next exiting thread, no need to gc to collect thread stacks anymore
901 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
902 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
903 ** bug fix: dynamic variable and thread start related gc lossage
904 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
906 * fixed some bugs revealed by Paul Dietz' test suite:
907 ** SUBTYPEP is slightly more accurate on heinously complicated
908 CONS types where some of the members have uncertain (in the
909 NIL, NIL sense) type relationships to each other.
911 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
912 no longer purified unless :PURIFY T is explicitly specified.
913 ** Non-purified cores are significantly smaller than before
915 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
916 * new feature: timers based on Zach Beane's excellent timer package
917 * added support for the following external formats: koi8-u,
918 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
919 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
920 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
921 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
922 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
923 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
924 * incompatible change: a threaded SBCL will no longer revert to
925 non-threaded mode on non-NPTL systems, but refuse to start entirely.
926 * bug fix: interrupts are disabled until startup is complete; no
927 more sigsegvs when receiving a signal to soon
928 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
930 * bug fix: add a workaround for the memory randomization features in
931 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
932 workaround will only be in effect on systems with the proc filesystem
934 * bug fix: printing objects of type HASH-TABLE signals a
935 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
937 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
938 have been running for over 50 days. (reported by Gilbert Baumann)
939 * bug fix: the logic for getting names of functions gets less
940 confused when confronded with alternate-metaclass
941 funcallable-instances. (reported by Cyrus Harmon)
942 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
943 consistent, even on internal alternate-metaclass objects.
944 * bug fix: flush closure information collected by physical
945 environment analysis prepass before the main pass. (bug reported
947 * bug fix: compiler pack phase does not modify a hashtable, which is
948 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
949 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
950 may be expanded. (reported by Alexander <malishev> on c.l.l)
951 * bug fix: no more style-warnings when compiling code using LAST
952 under high-SPEED low-SPACE settings. (reported by David Wallin
954 * fixed some incompatibilities between SBCL's MOP and the MOP
956 ** the METAOBJECT class is now implemented;
957 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
958 STANDARD-OBJECT, as required;
959 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
960 now compatible (as required by VALIDATE-SUPERCLASS); there
961 remains a consistency requirement over the presence or absence
962 of the FUNCTION class in the superclass list of finalized
963 classes; see the manual for more details;
964 ** the :AROUND method for COMPUTE-SLOTS on
965 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
966 requested slot ordering.
968 ** bug fix: parent thread now can be gc'ed even with a live
970 ** bug fix: binding a special with PROGV to no value is not
971 the same as that symbol not having been bound (thanks to
973 ** bug fix: binding specials is thread safe (thanks to
975 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
976 and SB-SPROF do not die with 'no handler for signal XX in
977 interrupt_handle_now(..)' anymore
978 ** bug fix: WITH-TIMEOUT works with multiple running threads
979 * fixed some bugs revealed by Paul Dietz' test suite:
980 ** ENSURE-GENERIC-FUNCTION should take a method class object for
981 the :method-class keyword argument.
983 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
984 * new port: the Solaris operating system on x86 processors is now
985 mostly supported, though some rough edges in the environment
986 remain. (thanks to Daisuke Homma)
987 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
988 space for its heap. (thanks to Thiemo Seufer)
989 * optimization: SBCL on MIPS platforms now supports dynamic-extent
990 closures. (thanks to Thiemo Seufer)
991 * minor incompatible change: eof selects abort in the debugger.
992 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
993 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
995 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
996 deprecated, and will go away in a future revision of SBCL.
997 * minor incompatible change: GC-ON and GC-OFF are no longer
998 implemented with a counter, it does not matter how many times gc
999 is switched on or off
1000 * bug fix: discriminating functions for generic function classes
1001 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1002 make invalid assumptions about method precedence order. (reported
1004 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1005 platforms. (thanks to Thiemo Seufer)
1006 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1007 * bug fix: correct pathname printing with printer escaping is on.
1008 (thanks to Kevin Reid)
1009 * bug fix: complex VOP definitions in "user-space" no longer trigger
1010 package locks. (reported by Zach Beane)
1011 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1012 for funcallable-instance functions, and is no different from
1014 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1015 characters in the current readtable. (reported by Nicholas Neuss)
1016 * bug fix: SUBTYPEP on various CONS types returns more a more
1017 accurate acknowledgment of its certainty.
1018 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1019 lists in some cases. This partially fixes bug 384.
1020 * flush all standard streams before prompting in the REPL and the
1022 * bug fix: signal handling and triggering gc do not conflict
1023 directly anymore, in particular a high frequency sb-sprof does
1024 not prevent gc from running
1025 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1026 approximation for timezone and DST information between the
1027 universal time epoch and the smallest negative 32-bit time_t.
1028 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1029 year 1899 so long as the encoded time is non-negative.
1030 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1031 the correct answer for non-constant rotate amounts when performing
1032 an inline 32-bit rotation.
1034 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1035 there is only one thread in the session
1036 ** bug fix: memory leak for streams created in one thread and
1037 written to in another
1038 ** bug fix: lockup when compiled with gcc4
1039 ** bug fix: race that allows the gc to be triggered when gc is
1041 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1042 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1044 ** bug fix: finished threads are gc'ed properly
1045 * fixed some bugs revealed by Paul Dietz' test suite:
1046 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1047 has arguments with a different set of applicable methods from
1048 the orignal arguments.
1049 ** The type error thrown by MAP now has a correct expected-type
1051 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1052 name a compiled function.
1053 ** Three MISC tests where a large bignum was improperly coerced to
1054 a float that couldn't represent that bignum during type
1055 derivation were fixed.
1056 ** SUBTYPEP can now handle types involving the negation of a
1057 list-form FUNCTION type.
1058 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1059 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1060 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1062 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1063 * New feature: Experimental support for bivalent streams: streams
1064 opened with :element-type :default now allow character and binary
1065 (unsigned-byte 8) I/O
1066 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1067 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1069 * bug fix: no more highly sporadic "couldn't check whether ~S is
1070 readable" when reading a stream and an interrupt hits in the middle
1071 of a select system call
1072 * compiler better recognizes complex arrays (reported by Eduardo
1074 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1075 for variables on SBCL built with linkage-tables. (reported by Luis
1077 * various error reporting improvements.
1078 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1079 (Thanks to Thiemo Seufer)
1080 * bug fix: nested reader invokations work correctly
1081 * bug fix: it is possible to have simultaneous references to foreign
1082 code and foreign data with the same name.
1084 ** added x86-64 support
1085 ** incompatible change: the threading api now works with thread
1086 objects instead of thread ids
1087 ** bug fix: threads are protected from signals and interruption when
1088 starting up or going down
1089 ** bug fix: a race where an exiting thread could lose its stack to gc
1090 ** bug fix: don't halt on infinite error in threads if possible
1091 ** fixed numerous gc deadlocks introduced in the pthread merge
1092 ** bug fix: fixed thread safety issues in read and print
1093 ** bug fix: debugger doesn't hang on session lock if interrupted at
1094 an inappropriate moment
1095 ** bug fix: run-program is now thread safe(r)
1096 ** bug fix: inner with-recursive-lock no longer releases the mutex
1097 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1098 * fixed some bugs revealed by Paul Dietz' test suite:
1099 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1100 EXPECTED-TYPE slots filled.
1101 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1103 ** MAKE-CONDITION accepts classes as type-designators.
1104 ** COMPILE may never return NIL.
1105 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1106 range before calling Unix time functions
1108 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1109 * numerous signal handling fixes to increase stability
1110 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1111 * minor incompatible change: we now correctly canonize default
1112 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1113 as per the MOP, rather than the historical (INITARG INITFUNCTION
1114 INITFORM). (reported by Bruno Haible)
1115 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1116 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1117 for more information.
1118 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1119 pathname is a directory pathname.
1120 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1121 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1123 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1124 :METACLASS argument in addition to a class name. (reported by
1125 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1126 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1127 (reported by Bruno Haible; patch from Gerd Moellmann)
1128 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1130 * bug fix: more cleanups to the floating point exception handling on
1131 x86-64 (thanks to James Knight)
1132 * bug fix: the compiler does not try to constant fold calls of
1133 COERCE and BIT-* functions when they return freshly constructed
1134 objects. (reported by Nikodemus Siivola and Paul Dietz)
1135 * optimization: improved the MIPS versions of generic (in the
1136 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1137 * optimization: direct conversion of (unsigned-byte 32) to floats on
1138 the PowerPC platform.
1139 * optimization: structure instances with raw slots now use less
1140 memory, and probably show better memory locality. (thanks to
1142 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1143 are treated as requests for stack allocation on the x86-64,
1144 Alpha-32, PPC and SPARC platforms.
1145 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1146 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1147 (thanks to Tony Martinez)
1149 ** gcing a dead thread can no longer lead to lockups
1150 ** threads block signals until they are set up properly
1151 ** errno is no longer shared by threads
1152 ** interrupt-thread restores the eflags register on x86
1153 ** fixed some lockups due to gc/thread interaction
1154 * fixed some bugs revealed by Paul Dietz' test suite:
1155 ** invalid dotted lists no longer raise a read error when
1156 *READ-SUPPRESS* is T
1157 ** signal an error if a symbol that names a declaration is used
1158 as the name of a type, or vice versa
1159 ** allow using the (declare (typespec var*)) abbreviation for
1160 (declare (type typespec var*)) with all type specifiers
1161 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1162 ** always return NIL from PPRINT-POP when OBJECT is NIL
1163 ** don't signal errors when pretty-printing malformed LABELS,
1164 FLET or MACROLET forms
1165 ** declarations in a DOLIST body are also in scope for the
1167 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1169 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1172 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1173 * fixed cross-compiler leakages that prevented building a 32-bit
1174 target with a 64-bit host compiler.
1175 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1176 opened with :IF-EXISTS :OVERWRITE.
1177 * fixed bug 281: error for an invalid qualifier in a short-form method
1178 combination method is not signalled until the faulty method is called.
1179 * bug fix: iteration variable type inferrer failed to deal with open
1180 intervals. (reported by Alan Shields)
1181 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1182 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1183 or more runtime options were provided to the sbcl binary.
1184 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1185 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1186 * optimization: there's now a fast-path for fixnum arguments in the
1187 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1188 * optimization: the code generated on x86-64 is more compact thanks
1189 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1190 * bug fix: floating-point exception handling now partly works on
1191 x86-64. (Thanks to James Knight)
1192 * improvement to the MIPS backend from Thiemo Seufer: C-style
1193 64-bit long long arguments and return values to alien functions
1195 * fixed some bugs revealed by Paul Dietz' test suite:
1196 ** the type-error signalled from WARN has a filled-in DATUM slot.
1197 ** the type-error required when a stream is not associated with
1198 a file has the stream as its datum.
1199 ** type-errors on single-floats on x86-64 no longer have
1200 :INVALID-OBJECT as the datum
1201 ** the type-errors signalled for invalid function names now have
1202 a correct expected type
1203 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1204 for structures defined with a :TYPE. Documentation strings for
1205 typed structures are no longer immediately discarded
1206 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1207 broadcast streams and synonym streams. FILE-LENGTH now also works
1208 on broadcast streams.
1210 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1211 * incompatible change: the --noprogrammer option, deprecated since
1212 version 0.7.5, has been removed. Please use the equivalent
1213 --disable-debugger option instead.
1214 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1216 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1217 inoperational for a while now) has been completely removed.
1218 * null lexical environments are now printed as #<NULL-LEXENV>,
1219 significantly reducing the amount of clutter in typical backtraces.
1220 * documentation on weak pointers, finalization, and after GC hooks
1221 has been added to the manual.
1222 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1223 as other specialized array types, is much faster. SUBSEQ and
1224 COPY-SEQ on such arrays have also been sped up.
1225 * optimization: EQL is now more efficient when at least other argument
1226 is known to be of type (OR FIXNUM (NOT NUMBER)).
1227 * fixed bug: the runtime is now less vulnerable to changes in the
1228 size of the SBCL object on OS X, and virtual memory is reserved for
1229 all spaces that need to be at a fixed address.
1230 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1231 * fixed bug: finalizers and after GC hooks that cause consing are now
1233 * fixed bug: compiler error messages and summaries are now printed to
1234 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1235 * fixed inference of the upper bound of an iteration variable.
1236 (reported by Rajat Datta).
1237 * fixed bug 211e: calling local functions with duplicated constant
1238 keyword argument no longer causes a bogus style warning about an
1240 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1241 ftype declaration to be disregarded. (reported by Dave Roberts)
1242 * fixed bug 373: caused by erronous compilation of references to alien
1243 variables in the runtime on ppc/darwin.
1244 * fixed bug 376: CONJUGATE type deriver.
1245 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1246 (reported by Baughn on #lisp)
1247 * a cross-compiler bug on non-x86 platforms has been identified and
1248 fixed. (thanks to Bruno Haible)
1249 * improvements to the MIPS runtime code for increased stability.
1250 (thanks to Thiemo Seufer)
1251 * increased the maximimum compact environment size to allow
1252 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1253 * partial workaround for bug 135: don't name the function we're
1254 calling for hairy cases of EVAL, so as not to accumulate one environment
1255 entry per call to EVAL. (reported by Kevin Reid)
1256 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1257 * optimization: type testing for non-vector arrays should be faster.
1258 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1260 * bug fix: setting 31st element of a bit vector to zero did not work
1262 * bug fix: redefining a class definition which failed due to a
1263 previous accessor / function clash now works (but see BUGS entry
1264 #380 for more problems in this area). (thanks to Zach Beane)
1265 * the long form of DEFINE-METHOD-COMBINATION disables method group
1266 checking when given a single method group with pattern *. (thanks
1268 * on x86 compiler supports stack allocation of results of simple
1269 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1270 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1271 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1272 * fixed some bugs related to Unicode integration:
1273 ** the restarts for recovering from input and output encoding
1274 errors only appear when there is in fact such an error to
1276 * fixed some bugs revealed by Paul Dietz' test suite:
1277 ** MISC.549 and similar: late transformation of unsafe type
1278 assertions into derived types caused unexpected code
1280 ** SCALE-FLOAT type deriver is less wrong.
1281 ** type derivers for EXP, LOG and similar functions compute result
1282 types for complex arguments better.
1283 ** (MISC.563) CONJUGATE type deriver works for very restricted
1285 ** out-of-line type testers for character strings are available.
1286 ** EQUAL compiler transform understands specialness of objects
1288 ** accessing double-floats stored on the stack now works on x86-64.
1289 ** debugger internals could sometimes create invalid lispobjs,
1290 resulting in GC crashes.
1291 ** MISC.548: type check weakening can convert required type into
1293 ** initialization forms for bindings are not in scope of free special
1296 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1297 * incompatible change: thread support for non-NPTL systems has
1298 been removed - locking is buggy and unreliable. A threaded
1299 SBCL build will now warn at startup and refuse to create
1300 new threads, unless futex support is detected in the kernel
1301 * incompatible change: the top level REPL now has only an ABORT
1302 restart associated with it, not TOPLEVEL and ABORT as it used to.
1303 TOP and TOPLEVEL are now available as debugger commands for
1304 returning to the top level.
1305 * incompatible change: forms evaluated in the REPL now use the
1306 global optimization policy.
1307 * incompatible change: user- and system-initialization files are
1308 no longer processed with LOAD, but by READ and EVAL; hence the
1309 global optimization policy, startup package, readtable, etc,
1311 * The .fasl file format number has been incremented because of
1312 various incompatible changes.
1313 * internal entry point details and argument counts no longer appear
1314 in backtraces unless explicitly requested by setting
1315 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1316 * built-in and standard functions no longer have names like "top
1317 level local call to FOO".
1318 * fixed bug 32: functions defined in non-null lexical environments
1319 now have more legible printed representation
1320 * fixed bug 33: functions defined in non-null lexical environemnts
1321 are now more amenable to inspection by INSPECT.
1322 * workaround for bug 354: XEPs no longer appear in backtraces unless
1323 explicitly requested.
1324 * fixed bug: receiving the signal which results from attempting to
1325 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1326 SIGBUS on other *BSDs) on architectures where the C stack is also
1327 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1328 to debugger confusion or wild instability. (reported by Cheuksan
1330 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1331 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1332 (reported by Lutz Euler)
1333 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1334 compiled code. (reported by Paul Dietz)
1335 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1336 specialized parameter to a method to something that is not TYPEP
1337 the specializer is now possible.
1338 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1339 face of package deletion.
1340 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1341 pathname. (reported by David Tolpin on c.l.l)
1342 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1343 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1344 STRING-STREAM and FILE-STREAM, respectively.
1345 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1346 than previously; in addition, its test suite is now run on build.
1347 (thanks to Luke Gorrie)
1348 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1349 * optimization: added a immediate representation for single-floats
1351 * fixed some bugs revealed by Paul Dietz' test suite:
1352 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1353 ** attempting to create a package with a colliding nickname causes
1354 correctable errors to be signalled.
1355 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1356 ** several x86-64 backend bugs related to sign-extension of immediate
1359 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1360 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1361 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1363 * fixed loading of multiply forward-referenced layouts.
1364 (thanks to Cheuksan Wang)
1365 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1366 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1367 (see COMPILE-FILE documentation for details of :PRINT option)
1368 * fixed bugs 19 and 317: fixed-format floating point printing is
1369 more accurate. This also fixes a bug reported by Adam Warner
1370 related to the ~@F format directive.
1371 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1373 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1374 dispatch table if the from-char is a dispatch macro character.
1375 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1376 by the wordsize no longer ignore the last word. (reported by Lutz
1378 * fixed bug in type checking of dynamic-extent variables. (reported
1380 * optimization: sequence traversal functions use their freedom to
1381 coerce function designators to functions.
1382 * optimization: code with many calls to CLOS methods specialized on
1383 CLOS classes has had redundant type checks removed.
1384 * optimization: added declarations to speed up operations that access
1385 the internal character database (for example STRING-UPCASE)
1386 * optimization: comparison operations between floats and sufficiently small
1387 fixnums no longer create extra rationals
1388 * fixed some bugs related to Unicode integration:
1389 ** portions of multibyte characters at the end of buffers for
1390 character-based file input are correctly transferred to the
1391 start of the buffer at the next read.
1392 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1393 passing it through to OPEN.
1394 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1395 argument given to internal calls to OPEN.
1396 ** the built-in comment readers (introduced by character sequences
1397 ";" and "#|") are more forgiving to encoding errors; they will
1398 STYLE-WARN and then attempt to resync the stream at a character
1399 boundary. (thanks to Teemu Kalvas)
1400 * fixed some bugs in the x86-64 port:
1401 ** Negative short int return values from c-calls are sign-extended
1403 ** The stack is aligned to 16-bytes for c-calls, as required by
1404 the ABI. (thanks to Cheuksan Wang)
1405 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1406 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1407 for enums. (thanks to Vincent Arkesteijn)
1408 ** Multiple small optimizations and bugfixes for floating point
1410 * fixed some bugs revealed by Paul Dietz' test suite:
1411 ** Space, Tab, Linefeed, Return and Page have the invalid
1412 secondary constituent character trait.
1413 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1415 ** WITH-INPUT-FROM-STRING should only update the index place on
1417 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1418 works more reliably.
1419 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1420 with one-dimensional arrays with fill pointers.
1421 ** TYPE-OF failed on a complex with an integer realpart and a
1423 ** compiler failure during type inference for the code of form
1424 (IF (EQL X (THE ...)) ...) (MISC.535).
1426 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1427 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1428 platform. The port supports 61-bit fixnums, large memory spaces
1429 and reloading shared object files.
1430 * enhancement: saving cores with foreign code loaded is now
1431 supported on ppc/Darwin in addition to the previously supported
1433 * enhancement: the statistical profiler now walks deeper into the
1434 call stack for more meaningful call-graphs and accrued time
1435 reports (x86/x86-64 only). It also now reports time spent in
1437 * enhancement: it is now possible to trace most individual methods
1438 of a generic function in addition to tracing the generic function
1440 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1442 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1443 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1444 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1445 produces an error. (thanks to Vincent Arkesteijn)
1446 * bug fix: NAMESTRING on pathnames with :WILD components in their
1447 directories works correctly. (thanks to Artem V. Andreev)
1448 * fixed bug 125: compiler preserves identity of closures. (reported
1450 * bug fixed: functions with &REST arguments sometimes failed with
1451 "Undefined function" when compiled with (DEBUG 3). (reported by
1453 * bug fix: overflow during compiling of setting element of a bit
1454 vector with constant index and value. (reported by Timmy Douglas)
1455 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1456 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1457 * on x86 compiler partially supports stack allocation of dynamic-extent
1459 * GO and RETURN-FROM do not check the extent of their exit points
1460 when compiled with SAFETY 0.
1461 * fixed some bugs related to Unicode integration:
1462 ** encoding and decoding errors are now much more robustly
1463 handled; it should now be possible to recover even from invalid
1464 input or output to the terminal. (thanks to Teemu Kalvas)
1465 ** provided a first cut at implementing STRING-TO-OCTETS and
1466 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1467 ** altered the SB-MD5 contributed module slightly, changing the
1468 interface just enough for it to be supportable for builds where
1469 lisp characters are not eight bits.
1470 * fixed some bugs revealed by Paul Dietz' test suite:
1471 ** the FORMATTER-generated functions for ~V[ conditionals require
1472 the correct number of arguments.
1473 ** READ-FROM-STRING returns the mandated second value when applied
1474 to displaced strings.
1475 ** the #\Rubout and #\Backspace characters are treated as invalid
1476 constituent characters by the tokenizer.
1478 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1479 * new feature: reloading changed shared object files with
1480 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1481 * new feature: references to foreign variables and functions
1482 can now be compiled and loaded before the corresponding shared
1483 object file is loaded, as long as the foreign definitions are
1484 available at runtime.
1485 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1486 supported. (thanks to Dan Debertin)
1487 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1488 just keywords. (thanks to Vincent Arkesteijn)
1489 * fixed bug #331: structure-class instances corresponding to
1490 DEFSTRUCT forms are now created eagerly.
1491 * fixed bug #345: backtraces from calls to undefined functions work
1492 on x86 as well. Related bug #61 is now also partially fixed on x86
1493 (backtraces from throws to unknown catch tags.)
1494 * bug fix: lambda-list parsing is now stricter vrt. order and number
1495 of lambda-list keywords.
1496 * bug fix: as specified by AMOP, an error is signalled if a
1497 class-option appears multiple times in a DEFCLASS form. (reported
1499 * bug fix: RANDOM can be compiled when the compiler derives the type
1500 of its numeric argument as a disjoint set of small integers.
1501 (reported by Paul Dietz)
1502 * bug fix: starting a core saved with shared objects loaded when
1503 those objects are not available no longer causes threaded SBCL to
1504 hang. (reported by Sean Ross)
1505 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1507 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1508 argument, which must always be NIL. (reported by Kalle Niemitalo)
1509 * bug fix: printing 1.0d+23 no longer results in an error.
1510 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1511 * bug fix: structure slot setters preserve evaluation order. (thanks
1513 * fixed some bugs related to Unicode integration:
1514 ** RUN-PROGRAM can allow its child to take input from a Lisp
1515 stream. (reported by Stefan Scholl)
1516 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1517 locales. (reported by Ken Causey)
1518 * fixed some bugs revealed by Paul Dietz' test suite:
1519 ** INCF, DECF and REMF evaluate their place form as specified in
1521 ** FORMATTER expands ~{ ~} iteration directives with V or #
1522 parameters correctly.
1523 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1524 consequent uses no arguments correctly.
1525 ** the system has a partial understanding of the (COMPLEX RATIO)
1527 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1530 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1531 * new feature: a build-time option (controlled by the :SB-UNICODE
1532 keyword feature, enabled by default) for building the system with
1533 support for the entire 21-bit character space defined by the
1535 * new feature: the system now has rudimentary external-format
1536 support; the primary user-visible change at this time is that
1537 characters with the high bit set (such as lower-case-e-acute) will
1538 print correctly to a terminal in a UTF-8 environment.
1539 * minor incompatible change: BASE-CHAR no longer names a class;
1540 however, CHARACTER continues to do so, as required by ANSI.
1541 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1542 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1543 has been moved to the SB-EXT package (temporarily re-exported from
1545 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1546 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1548 * fixed bug #351: better error-handlind and reporting for malformed
1550 * fixed bug #350: bignum-printing is now more memory-efficient,
1551 allowing printing of very large bignums, eg. (expt 2 10000000).
1552 (reported by Bruno Haible)
1553 * fixed bug #302: better primitive-type selection for intersection
1555 * fixed bug #308: non-graphic characters now all have names, as
1556 required. (reported by Bruno Haible)
1557 * bug fix: redefining a class with different superclasses now correctly
1558 removes it from the direct-subclasses of its previous superclasses.
1559 (reported by David Morse)
1560 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1561 new value now works. (reported by Bruno Haible)
1562 * bug fix: correct canonicalization of multiple non-standard slot
1563 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1564 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1565 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1566 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1567 now exists, an signals an error.
1568 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1569 during macroexpansion for non-iteration variables as well. (reported
1570 by Bruno Haible for CMUCL)
1571 * bug fix: Cyclic structures and unprintable objects in compiler
1572 messages no longer cause errors. (reported by Bruno Haible)
1573 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1574 and READ-FROM-STRING all now return a primary value of NIL if
1575 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1576 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1577 (reported by Bruno Haible for CMUCL)
1578 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1579 do not designate a condition. (reported by Bruno Haible for
1581 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1582 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1583 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1584 permitting resolution in favour of any of the conflicting symbols.
1585 (reported by Bruno Haible for CMUCL)
1586 * bug fix: EQUAL compiler optimizations is less aggressive on
1587 strings which can potentially compare true despite having distinct
1588 specialized array element types.
1589 * bug fix: unit enumerations can be defined without dividing by
1590 zero. (thanks to Vincent Arkesteijn)
1591 * FORMAT compile-time argument count checking has been enhanced.
1592 (report from Bruno Haible for CMUCL)
1593 * a partial workaround for the bug 262: the compiler does not try to
1594 inline-expand a local function doing RETURN-FROM from a deleted
1595 BLOCK. (thanks to Peter Denno for the bug report and to David
1596 Wragg for the simple test case)
1597 * fixed some bugs revealed by Paul Dietz' test suite:
1598 ** RENAME-PACKAGE allows all package designators as new package
1600 ** constraint propagation and lambda variable substitution are
1601 more cautious in dealing with partially deleted code.
1602 ** compiler sometimes lost reoptimization passes.
1603 ** CERROR, given a condition as condition designator, passes its
1604 remaining arguments to the continue format control without
1606 ** Case-altering FORMAT directives work correctly on non-ASCII
1608 ** The REAL type specifier handles bounds outside the floating
1609 point ranges without signalling FLOATING-POINT-OVERFLOW.
1610 ** Functions with IR1-transformations can create intercomponent
1611 references to global functions.
1612 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1614 ** FORMAT ~R treats a nil value for its first parameter correctly.
1616 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1617 * enhancement: saving cores with foreign code loaded is now
1618 supported on x86/NetBSD and sparc/Linux in addition to the previously
1619 supported platforms.
1620 * bug fix: on some platforms repeated installations caused multiple
1621 copies of HTML documentation to be installed -- should not happen
1622 any more. (reported by Stefan Scholl)
1623 * bug fix: parsing self-recursive alien record types multiple times
1624 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1625 original patch by Helmut Eller for CMUCL)
1626 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1627 (thanks to Richard Kreuter)
1628 * bug fix: defining classes whose accessors are methods on existing
1629 generic functions in other (locked) packages no longer signals
1630 bogus package lock violations. (reported by François-René Rideau)
1631 * bug fix: special variables as DEFMETHOD parameters no longer have
1632 associated bogus type declarations. (reported by David Wragg and
1634 * bug fix: read-write consistency on streams of element-type
1635 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1636 * bug fix: redefiniton of the only method of a generic function with
1637 no DEFGENERIC no longer emits a full WARNING. In addition,
1638 redefinition of generic functions with no DEFGENERIC to an
1639 incompatible lambda list now signals an error. (thanks to Zach
1641 * bug fix: DEFGENERIC now works even when there's a function of the
1642 same name in an enclosing lexical environment. (thanks to Zach
1644 * fixed compiler failure, caused by instrumenting code during
1645 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1646 * optimization: added loop analysis and improved register allocation
1647 * fixed some bugs revealed by Paul Dietz' test suite:
1648 ** POSITION on displaced vectors with non-zero displacement
1649 returns the right answer.
1650 ** (SIMPLE-STRING) is a valid type specifier for sequence
1652 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1654 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1655 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1657 ** The justification version of the FORMAT ~< directive treats
1658 non-zero minpad parameter correctly.
1660 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1661 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1662 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1663 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1664 the supported interface.
1665 * new feature: Single-stepping of code compiled with DEBUG 2 or
1666 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1667 * new feature: saving cores with foreign code loaded is now
1668 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1669 Timothy Moore's work for CMUCL)
1670 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1671 parameters to * instead of NIL if no initform is supplied.
1672 (reported by Johan Bockgaard)
1673 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1674 correctly when FUNCALL appears as the car of the form. Note:
1675 despite this FUNCALL forms are not currently subject to
1676 compiler-macro expansion. (port of Raymond Toy's fix for the
1677 same from CMUCL, reported by Johan Bockgaard)
1678 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1679 (thanks for Teemu Kalvas)
1680 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1681 case instead of terminating the iteration (thanks for Julian
1682 Squires, Sean Champ and Raymond Toy)
1683 * bug fix: incorrect expansion of defgeneric that caused a style
1684 warning. (thanks for Zach Beane)
1685 * bug fix: slot accessor effective method computation works properly
1686 for classes with multiple non-standard applicable methods on
1687 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1688 * on x86 compiler supports stack allocation of results of LIST and
1689 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1690 CMUCL implementation by Gerd Moellmann)
1691 * fixed some bugs revealed by Paul Dietz' test suite:
1692 ** FORMAT strings with both the ~<~:;~> form of the justification
1693 directive and pretty-printing directives cause an error.
1695 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1696 * incompatible change: the internal functions
1697 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1698 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1699 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1700 instead of the old functions.
1701 * new feature: on platforms where "dladdr" is available foreign
1702 function names now appear in backtraces. (based on Helmut Eller's
1704 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1705 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1707 * documentation: ASDF manual is now distributed with SBCL.
1708 * bug fix: SBCL can now load its contributed modules with REQUIRE
1709 even if the system-provided entries have been removed from
1710 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1712 * bug fix: backtraces involving undefined functions or assembly
1713 routines are more informative. (thanks to Brian Downing)
1714 * bug fix: mutually referent alien structures now work correctly.
1715 (reported by Rick Taube)
1716 * bug fix: structures defined by WITH-ALIEN can be referred to
1717 within other definitions in the same WITH-ALIEN.
1718 * bug fix: division operators (MOD, TRUNCATE and the like) with
1719 constant zero divisors and integer dividends no longer generate
1721 * bug fix: provide default methods for INPUT-STREAM-P and
1722 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1723 * bug fix: improve the detection and resolution of MOP metacycles
1724 (where effective method computation depends on the generic function
1725 having its effective method computed). (reported by Bruno Haible)
1726 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1727 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1728 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1729 * bug fix: pathnames with relative directory components can be
1730 represented relative to default pathnames.
1731 * optimization: in taking the GCD of bignums, reduce the two bignums
1732 to approximately the same size (using Euclid's algorithm) before
1733 applying the more sophisticated binary GCD. (thanks to Juho
1735 * optimization: COUNT on bitvectors now operates word-at-a-time.
1736 * optimization: ASH with a positive, but not necessarily constant,
1737 (leftwards) shift, when in a modular context, is compiled to a
1739 * fixed some bugs revealed by Paul Dietz' test suite:
1740 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1741 if the corresponding argument is NIL.
1742 ** FORMAT directives accepting a minpad parameter treat negative
1744 ** literal commas as character parameters to FORMAT directives are
1746 ** literal spaces directly after ~<Newline> directives within a
1747 format-logical-block (~:< ~@:>) do not induce :FILL-style
1748 conditional newlines.
1749 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1750 as specified: it no longer includes conditional newlines).
1751 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1753 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1754 :PER-LINE-PREFIX argument does not evaluate to a string.
1756 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1757 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1758 the manual for details; for now, package locks can be disabled by
1759 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1760 if no major problems are found then it is likely that they will be
1761 compiled in unconditionally.
1762 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1763 unsupported operators on all platforms. To load a shared library
1764 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1765 object file, link it into a shared library outside of SBCL and
1766 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1768 * bug workaround: ROOM T and the GC were not working together
1769 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1770 reporting facility aren't preserved. That reporting has been
1771 disabled, so now until and unless someone figures out how to make it
1772 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1773 * minor incompatible change: as threatened around sbcl-0.8.0, the
1774 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1775 The symbols are also exported from SB-PCL for backwards
1776 compatibility, but more so than before SB-PCL should be treated as
1777 an implementation-internal package.
1778 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1780 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1781 It is known as of this release not to work on the Alpha, however.
1782 * fixed bug #167: errors signalled due to illegal syntax in method
1783 bodies are now more legible.
1784 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1785 designators and can hence be used with TYPEP.
1786 * fixed bug #333: CHECK-TYPE now ensures that the type error
1787 signalled, if any, has the right object to be accessed by
1788 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1789 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1790 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1792 * fixed bug #334: programmatic addition of slots using specialized
1793 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1794 and :ALLOCATION :CLASS slots.
1795 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1796 float exponents. (rereported by Peter Seibel)
1797 * fixed a bug: #\Space (and other whitespace characters) are no
1798 longer considered to be macro characters in standard syntax by
1799 GET-MACRO-CHARACTER.
1800 * fixed bug: initialization of condition class metaobjects no longer
1801 causes an instance of the condition to be created. (reported by
1803 * fixed bug: it is now possible to have slots such that
1804 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1805 description is neither :INSTANCE nor :CLASS.
1806 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1807 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1808 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1809 non-standard SLOT-DEFINITION-ALLOCATION values.
1810 * fixed bugs: various race conditions handling exiting threads.
1811 CL-PPCRE's thread torture test now passes.
1812 * fixed arguably-a-bug: GC time proportional to number of threads in
1813 system even when most of them are idle
1814 * optimization: improved performance of BIT and SBIT on bit-vectors.
1815 * .fasl file incompatibility: The fasl file version number has
1816 been incremented because of changes associated with package locks.
1818 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1819 * minor incompatible change: the system no longer provides
1820 optimization hints (conditions which are TYPEP
1821 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1822 mode; these hints are still emitted when higher SPEED optimization
1824 * new contrib module: a sampling profiler (profiling by statistical
1825 sampling, rather than by instrumenting functions) is available as
1826 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1827 * the behaviour of the standard function ED is now customizeable by
1828 third parties through a hook variable: see ED's documentation
1829 string for information on the protocol.
1830 * the compiler no longer emits efficiency notes for (FUNCALL X)
1831 when the type of X is uncertain under default optimization
1833 * fixed bug 276: mutating a binding of a specialized parameter to a
1834 method to something that is not TYPEP the specializer is now
1836 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1837 exhibit the required behaviour on the x86 platform. (thanks to
1838 Peter van Eynde, Eric Marsden and Bruno Haible)
1839 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1840 even for difficult arguments. (reported by Peter Graves)
1841 * fixed bug 141a: the backquote printer now descends quoted
1843 * fixed another bug in backquote printing: no more destructive
1844 modification of the form's list structure. (reported by Brian
1846 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1847 move between its address being taken and the call to
1848 interrupt_thread, fixing a crashing race condition.
1849 * the SB-POSIX contrib implementation has been adjusted so that it
1850 no longer exhibits ridiculously poor performance when constructing
1851 instances corresponding to C structs.
1853 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1854 * minor incompatible change: the sb-grovel contrib now treats C
1855 structures as alien (in the sense of SB-ALIEN) objects rather than
1856 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1857 has implications for memory management of client code
1858 (sb-grovel-returned objects must now be manually managed) and for
1859 type safety (alien objects now have full types).
1860 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1861 used to control emission of compiler diagnostics, rather than the
1862 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1863 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1864 quality should be considered deprecated.
1865 * install.sh now installs the user manual as well
1866 * (not quite a new documentable feature, but worth considering in
1867 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1868 beginnings of a semantically meaningful condition hierarchy is
1869 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1871 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1872 designator as the defaults argument.
1873 * fixed bug: Displaced arrays whose displaced-to array has become
1874 too small now cause ARRAY-DIMENSION to signal an error, providing
1875 for safer bounds-checking. (reported by Bruno Haible)
1876 * fixed bug: DEFCLASS slot definitions with identical :READER and
1877 :WRITER names now signal a reasonable error. (reported by Thomas
1879 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1881 * fixed bug: Invalid slot specification errors now print correctly.
1882 (thanks to Zach Beane)
1883 * fixed bug 320: Shared to local slot value transfers in class
1884 redefinitions now happen corrently from superclasses as
1885 well. (reported by Bruno Haible)
1886 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1888 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1889 improper lists correctly. (reported by Bruno Haible)
1890 * fixed bug 313: source-transform for <fun-name> was erroneously
1891 applied to a call of a value of a variable with name <fun-name>.
1892 (reported by Antonio Menezes Leitao)
1893 * fixed bug 307: The obsolete instance protocol ensures that
1894 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1895 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1896 remove dead unknown-values globs from the middle of the stack.
1897 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1898 easier to resume long computations after using *BREAK-ON-SIGNALS*
1899 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1900 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1901 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1902 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1903 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1904 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1906 * fixed a (fairly theoretical) bug in string printing: if
1907 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1908 does not have array-element-type equal to the most general string
1910 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1911 function name. (thanks to Nikodemus Siivola)
1912 * fixed bug: compile-time format string checker failed on
1913 non-closed ~{. (reported by Thomas F Burdick)
1914 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1915 don't warn on legal loop code involving more than one aggregate
1916 boolean. (thanks to Nikodemus Siivola)
1917 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1918 in keyword position to MAKE-PACKAGE was not being reported as
1920 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1921 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1923 * fixed bugs 280 and 312: the checking for multiple definitions in a
1924 file is less likely to become confused by uses of inline
1926 * fixed bug: the #S reader macro performs the keyword coercion
1927 specified for slot names. (reported by Kalle Niemitalo)
1928 * fixed bug: lambda lists may contain symbols whose names start with
1929 & but are not lambda-list-keywords; their occurrence triggers a
1931 * fixed bug 321: define-method-combination argument lambda lists do
1932 not cause invalid code to be generated when &optional or &aux
1933 variables have default values. (reported by Bruno Haible)
1934 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1935 have CLOS classes; this appears also to have fixed reported
1936 instability in stack exhaustion detection.
1937 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1938 MUFFLE-WARNING when no associated restart is present is now
1940 * optimization: rearranged the expansion of various defining macros
1941 so that each expands into only one top-level form in a
1942 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1944 * optimization: used a previously little-used slot in symbols to
1945 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1947 * fixed some bugs revealed by Paul Dietz' test suite:
1948 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1950 ** Fixed an optimization bug related to inheritance of initforms
1951 from local to shared slots.
1952 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1953 ** CAST optimizer forgot to flush argument derived type.
1954 ** print/read consistency on floats is now orders of magnitude
1955 more likely. (thanks also to Bruno Haible for a similar report
1957 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1958 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1959 for objects of type REAL. Make it so.
1960 ** SXHASH is specified (infelicitously) to respect similarity,
1961 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1962 it so. (thanks to Markus Ziegler)
1963 ** on the Alpha, the compiler succeeds in compiling functions
1964 returning a known number of arguments greater than 63.
1965 ** fixed handling of invalid NIL arguments in keyword position
1967 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1969 ** provided a readably-printable representation for RANDOM-STATE
1971 ** ensured that pathnames, where they have a namestring, always
1972 print using #P"..." syntax.
1974 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1975 * Support for the forthcoming 2.0 version of the NetBSD kernel
1976 running on x86 hardware has been added. (thanks to Perry
1977 E. Metzger most immediately, and others for their past work)
1978 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1979 current release version; SBCL's OpenBSD support had been broken
1980 since about the time of OpenBSD's switch to ELF binary format.)
1981 * [placeholder for DX summary]
1982 ** user code with &REST lists declared dynamic-extent, under high
1983 speed or space and low safety and debug optimization policy.
1984 * The manual has been converted to Texinfo format and the debugger
1985 chapter from the cmucl manual has been added.
1986 * A facility has been added to extract documentation strings from
1987 sbcl and store them as Texinfo-formatted snippets for inclusion in
1988 the manual (via Texinfo's @include directive)
1989 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1990 the test case to Dave Roberts)
1991 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1992 pointers. (reported by Sean Ross)
1993 * bug fix: PROFILE output is printed nicely even for large numerical
1994 values. (thanks to Zach Beane)
1995 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1996 greater than 32 handle EOF correctly.
1997 * bug fix: on X86 an immediate argument of the IMUL instruction is
1998 correctly printed in disassembly. (thanks to Lutz Euler)
1999 * bug fix: class slots in redefined classes preserve their old
2000 values. (thanks to Bruno Haible and Nikodemus Siivola)
2001 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2002 succeeds. (reported by Marco Baringer)
2003 * fixed some bugs revealed by Paul Dietz' test suite:
2004 ** READ-SEQUENCE now works on ECHO-STREAMs.
2005 ** RATIONALIZE works more according to its specification. (thanks
2008 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2009 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2010 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2011 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2012 should matter to you only if you rebind the printer control
2013 variables and then find you want different bindings in the
2014 debugger than in the ordinary execution of your program.)
2015 * The runtime build system has been tweaked to support building
2016 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2017 assembler and linker. (thanks to Nikodemus Siivola)
2018 * Unbound, undefined, undeclared variables now trigger full
2019 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2020 more often programmer error than deliberate exploitation of undefined
2022 * optimization: the hash algorithm for strings has changed to one
2023 that is less vulnerable to spurious collisions. (thanks to Juho
2025 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2026 needless bounds checking. (thanks to Juho Snellman)
2027 * optimization: implemented multiplication as a modular
2028 (UNSIGNED-BYTE 32) operation on the PPC backend.
2029 * fixed some bugs revealed by Paul Dietz' test suite:
2030 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2032 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2033 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2034 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2036 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2038 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2040 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2041 their output stream on EOF from read.
2042 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2043 have been read to end-of-file.
2044 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2046 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2047 description of determination of which consecutive characters
2049 ** Printing the "Space" character with escaping on now yields "#\\ ",
2050 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2051 ** Reading floating-point numbers with *READ-BASE* set to a number
2052 less than 10 works correctly.
2053 ** Reading floating-point numbers with *READ-BASE* set to a number
2054 more than 10 works correctly.
2055 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2056 the readtable currently in effect.
2058 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2059 * minor incompatible change: parsing of namestrings on a physical
2060 (Unix) host has changed; numbers after the final #\. in a
2061 namestring are no longer interpreted as a version field. This is
2062 intented to be largely invisible to the user, except that the
2063 meaning of the namestring "*.*.*" has changed: it now refers to a
2064 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2065 should usually be replaced by
2066 (make-pathname :name :wild :type :wild :version :wild)
2067 with the added benefit that this is more likely to be portable.
2068 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2069 option to OPEN now signals an error if the file being opened
2070 exists; this may have an impact on existing code.
2071 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2072 signals received, on the PowerPC platforms (both Linux and
2073 Darwin). (thanks to Pierre Mai for pointing out the location of
2075 * several fixes on OS X: The system now builds and runs cleanly on
2076 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2077 (thanks to Brian Mastenbrook)
2078 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2079 second-resolution: integer multiples of 1/3600 between -24 and 24.
2080 (thanks to Vincent Arkesteijn)
2081 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2082 type when called with 1 argument; PEEK-CHAR checked type of
2083 PEEK-TYPE only after having read first character from a
2084 stream. (reported by Peter Graves)
2085 * bug fix: the garbage collector now has much better locality
2086 behaviour, and in particular no longer treats all memory as being
2087 exhausted when many small objects point to each other in a deeply
2089 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2090 recognized as being TYPEP their class.
2091 * bug fix: the PUSHNEW documentation string has been corrected.
2092 (thanks to Vincent Arkesteijn)
2093 * bug fix: defaulting of the value for the last of an atypically
2094 large number of multiple values being bound was not being
2095 performed correctly on the Alpha or PPC platforms
2096 * optimization: implemented multiplication as a modular
2097 (UNSIGNED-BYTE 32) operation on the x86 backend.
2098 * optimization: SEARCH on simple-base-strings can now be open-coded.
2099 (see also contrib/compiler-extras.lisp for inspiration for
2100 teaching the compiler about the Boyer-Moore algorithm).
2101 * value, returned by MAX (and MIN) called with several EQUALP, but
2102 not EQL, arguments now does not depend on compiler settings.
2103 * fixed some bugs revealed by Paul Dietz' test suite:
2104 ** in stack analysis liveness information is propagated from
2105 non-local entry points.
2106 ** pathwise CAST removing failed when the CAST node did not start
2108 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2109 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2111 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2112 host is already defined.
2113 ** RENAME-FILE works on streams instead of signalling an internal
2115 ** PEEK-CHAR uses the current readtable when determining whether
2116 or not a character is whitespace.
2117 ** MERGE-PATHNAMES handles the case when the pathname does not
2118 specify a name while the default-pathname specifies a version.
2119 ** Pathnames now stand a better chance of respecting print/read
2121 ** Attempting to use standardized file system operators with a
2122 pathname with invalid :DIRECTORY components signals a
2124 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2126 ** DIRECTORY on logical pathnames is more correct.
2127 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2128 signal a TYPE-ERROR if their argument is not a stream.
2129 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2130 designator argument does not designate a stream.
2131 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2132 examining the synonym.
2133 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2135 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2136 element-type larger than ([UN]SIGNED-BYTE 32).
2138 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2139 * When built with the :SB-FUTEX feature, threaded builds now take
2140 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2141 for faster/more reliable mutex and condition variable support.
2142 * Incompatible change (but one you probably shouldn't have been using
2143 anyway): the interface and code for arbitrating between multiple
2144 threads in the same user session has been redesigned.
2145 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2146 error when called without an explicit environment argument.
2147 (thanks to Nikodemus Siivola)
2148 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2149 confused about their position. (thanks to Adam Warner and Gerd
2151 * bug fix: Pretty printing backquoted forms with unquotations in the
2152 argument list position of various code constructs such as LAMBDA
2153 now works correctly. (reported by Paul Dietz)
2154 * bug fix: Pretty printing unquotations no longer loses all
2155 stream position information.
2156 * optimization: performance of string output streams is now less
2157 poor for multiple small sequence writes.
2158 * optimization: performance of CSUBTYPEP in the presence of complex
2159 expressions involving CONS and NOT many times has been improved.
2160 (reported by Paul Dietz)
2161 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2163 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2164 streams for the REPL, for the debugger, and so on. (thanks to
2166 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2167 * fixed some bugs revealed by Paul Dietz' test suite:
2168 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2169 is ignored if it is too small, rather than propagating through
2170 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2171 ** extremely complex negations of CONS types were not being
2172 sufficiently canonicalized, leading to inconsistencies in
2174 ** VALUES tranformer lost derived type.
2176 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2177 * fixed a bootstrapping bug: the build process no longer assumes
2178 that the various BOOLE-related constants have the same value in
2179 host and target lisps. (noted by Paul Dietz' test suite on an
2180 SBCL binary built from CLISP)
2181 * The system can now be dynamically linked on the MIPS platform,
2182 which enables dynamic loading of foreign code from Lisp. (thanks
2183 to Ralf Baechle for discussions on the MIPS ABI)
2184 * The system now records debugging information for its own source
2185 files in a filesystem-position-independent manner, relative to
2186 the "SYS" logical host.
2187 * fixed a compiler bug: MV-LET convertion did not check references
2188 to the "max args" entry point. (reported by Brian Downing)
2189 * tweaked disassembly notes to be less confident about proclaiming
2190 some instruction as an LRA. (thanks to Brian Downing)
2191 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2192 now each have their own history, command character, and other
2193 characteristics. (thanks to David Lichteblau)
2194 * fixed some bugs revealed by Paul Dietz' test suite:
2195 ** compiler failure in compiling LOGAND expressions including a
2197 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2198 shift greater than 32.
2199 ** FLUSH-DEST did not mark blocks for type check regeneration.
2200 ** HANDLER-CASE failed to accept declarations in handler clauses
2201 in some circumstances.
2203 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2204 * New code in contrib/sb-introspect (still probably not entirely
2205 stable yet) provides some support for smart Lisp development
2206 environments like SLIME.
2207 * The conditions signalled for errors occurring when loading .fasl
2208 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2209 in a way which should help ASDF recover gracefully.
2210 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2211 stuff has been cleaned up. If you code contrib/ stuff, this might
2212 affect you, and you can look at contrib/README, contrib/STANDARDS,
2213 and/or the 0.8.4.27 diff to check.
2214 * In full calls the compiler now does not generate checks for declared
2215 argument types for all arguments.
2216 * various threading fixes
2217 ** and some experimental patches which didn't make it into
2218 the main tree for this release, but which are shipped in
2219 contrib/experimental-thread.patch as a possible fix for some
2220 failures (deadlock, spinning...) in GC-intensive multithreaded
2222 * fixed PPC build problem (source code incompatibility of different
2223 library versions): added offsetof() hackery which attempts to divine
2224 where glibc maintainers put uc_mcontext today
2225 * fixed bug 282: compiler does not trust type assertions while passing
2226 arguments to a full call.
2227 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2228 &OPTIONAL VALUES type parameter.
2229 * fix bug 214: algorithm for noting rejected templates is now more
2230 similar to that of template seletion. (also reported by rydis on
2232 * fixed bug 141b: printing backquoted information readably and prettily
2233 inserts a space where necessary.
2234 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2235 (CHARACTER 10) are now reported as errors, rather than propagated
2236 as unknown types. (reported by piso on #lisp)
2237 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2238 with values NIL and :ERROR. (thanks to Milan Zamazal)
2239 * fixed bug 191c: CLOS now does proper keyword argument checking as
2240 described in CLHS 7.6.5 and 7.6.5.1.
2241 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2242 longer raise an error; further, using a list as a for-as-arithmetic
2243 counter now raises a meaningful error.
2244 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2245 now understood by sequence creation functions such as MAKE-SEQUENCE
2247 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2248 a STRING-INPUT-STREAM.
2249 * compiler enhancement: SIGNUM is now better able to derive the type
2251 * type declarations inside WITH-SLOTS are checked. (reported by
2253 * fixed some bugs revealed by Paul Dietz' test suite:
2254 ** incorrect optimization of TRUNCATE for a positive first
2255 argument and negative second.
2256 ** compiler failure in let-convertion during flushing dead code.
2257 ** compiler failure while deriving type of TRUNCATE on an
2258 interval, containing 0.
2259 ** ASH of a negative bignum by a negative bignum count now returns
2261 ** intersection of CONS types now canonicalizes properly, fixing
2262 inconsistencies in SUBTYPEP.
2264 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2265 * incompatible change: The --disable-debugger command line
2266 option now clobbers the debugger at a more fundamental
2267 level, by redefining #'INVOKE-DEBUGGER instead of by
2268 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2269 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2270 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2271 enter the debugger REPL and then suffer recursive errors
2272 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2273 while under the new system, BREAK in --disable-debugger mode
2274 terminates the system just as an unhandled error would.
2275 * fixed compiler performance when processing loops with a step >1;
2276 * bug fix: DOCUMENTATION now retrieves generic function
2277 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2278 support has been systematized, and now supports the methods
2279 specified by ANSI, along with a default method and a method for
2280 slot documentation. (reported by Nathan Froyd)
2281 * bug fix: effective methods associated with a generic function are
2282 no longer cached over a change of that generic function's method
2283 combination. (reported by Andreas Fuchs)
2284 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2285 names a non-existent directory. (thanks to Andreas Fuchs)
2286 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2287 the PPC platform, lead to stack corruption; this has been fixed.
2288 (reported by Rainer Joswig)
2289 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2290 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2291 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2292 arithmetic in general, is now much more reliable on the PPC
2294 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2295 platform now returns the right answer.
2296 * optimization: restored some effective method precomputation in
2297 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2298 precomputation is now tunable.
2299 * optimization: compiler-internal data structure use has been
2300 reviewed, and changes have been made that should improve the
2301 performance of the compiler by about 20%.
2302 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2303 simple-base-strings and simple-bit-vectors is improved.
2304 * optimization: the optimization of 32-bit logical and arithmetic
2305 functions introduced in version 0.8.3 on the x86 has been
2306 implemented on the mips, ppc and sparc platforms; an
2307 implementation of the same facility, but for 64-bit arithmetic,
2308 has been added for the alpha.
2309 * microoptimization: the compiler is better able to make use of the
2310 x86 LEA instruction for multiplication by constants.
2311 * bug fix: in some situations compiler did not report usage of
2312 generic arithmetic in (SPEED 3) policy.
2313 * bug 145b fix: compiler used wrong type specifier while converting
2314 MEMBER-types to numeric.
2315 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2317 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2318 index argument now works properly on the Alpha platform.
2319 * bug fix: floating point exception treatment on the Alpha platform
2321 * bug fix: FILE-POSITION works much better on string input and
2322 output streams. (thanks to Nikodemus Siivola)
2323 * bug fix: many threading/garbage collection symptoms sorted.
2324 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2325 might be pseudo-atomic.
2326 * internal change: Stopping for GC is now done with signals not
2327 ptrace. GC is now done in whichever thread wanted it, instead of
2329 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2331 * bug fix: result form in DO is not contained in the implicit
2333 * incompatible change: ICR structure is changed; the value part of
2334 CONTINUATION is now called LVAR; corresponding functions are
2335 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2336 * added type deriver for ISQRT (thanks to Robert E. Brown).
2337 * bug fix: better support for loading from the command line when an
2338 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2340 * fixed some bugs revealed by Paul Dietz' test suite:
2341 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2342 ** ROUND and FROUND now give the right answer when given very
2343 small float arguments.
2344 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2346 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2347 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2348 ** LCM should return a non-negative integer.
2349 ** PARSE-INTEGER returned the index of a terminator instead of the
2350 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2351 ** PARSE-INTEGER returned an incorrect index being applied to a
2353 ** LCM with two arguments of 0 returns 0 rather than signalling
2355 ** unsigned addition of a 32-bit constant with the high bit set no
2356 longer causes an internal compiler error on the x86.
2357 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2358 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2359 with negative last argument.
2360 ** byte specifiers with zero size and position no longer cause
2361 an error during type derivation.
2362 ** bignum multiplication on the Alpha platform now returns the
2364 * porting: The system now builds on SuSE AMD64, although it still
2365 generates a 32-bit binary.
2366 * .fasl file incompatibility: The fasl file version number has
2367 been incremented (because of the changes to internal compiler
2368 data structures referred to above).
2370 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2371 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2372 more accurately, on the Darwin kernel running on PowerPC hardware.
2373 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2374 * Compiler code deletion notes now signal a condition of type
2375 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2376 an associated MUFFLE-WARNING restart.
2377 * The compiler now performs limited argument count validation of
2378 constant format strings in FORMAT, and where appropriate in ERROR,
2379 CERROR and WARN. (thanks to Gerd Moellmann)
2380 * New ASDF-INSTALL contrib can be used for automatic download and
2381 installation of third-party Lisp code from CCLAN or other sites
2383 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2384 which forces another thread to execute a function supplied by the
2386 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2387 now accept and act upon their :ELEMENT-TYPE keyword argument.
2388 (reported by Martin Atzmueller, Edi Weitz)
2389 * bug fix: FILE-POSITION now accepts position designators up to
2390 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2391 is the greater. (thanks to Patrik Nordebo)
2392 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2393 Kalle Olavi Niemitalo)
2394 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2395 printable. (reported by Eric Marsden)
2396 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2397 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2398 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2399 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2400 (thanks to Stig Sandoe)
2401 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2402 UNREAD-CHAR now correctly includes the unread character in the
2403 target sequence. (thanks to Gerd Moellmann)
2404 * bug fix in threaded builds: the system can now be suspended and
2405 resumed by shell job control with minimal disruption.
2406 * bug fixes in times and timezones >2038AD
2407 * better handling of "where is GNU make?" problem in build scripts
2408 (thanks to Nikodemus Siivola)
2409 * new optimization: inside a named function any reference to a
2410 function with the same name is considered to be a self-reference;
2411 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2412 optimization quality.
2413 * new optimization on x86: logical functions and + now have
2414 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2415 used when the result is truncated to 32 bits.
2416 * VALUES declaration is partially enabled.
2417 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2418 * fixed some bugs revealed by Paul Dietz' test suite:
2419 ** The system now obeys the constraint imposed by
2420 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2421 types form a lattice under type intersection.
2422 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2423 ** ASSOC now ignores NIL elements in an alist.
2424 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2425 and (1+ MOST-POSITIVE-FIXNUM) answers.
2426 ** The addition of a method with invalid qualifiers to a generic
2427 function does not cause an error to be signalled immediately;
2428 a warning is signalled, and the error is generated only on
2429 calling the generic function.
2430 * changed .fasl file version number, in part to add type codes for
2431 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2432 obscure ANSI requirements
2434 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2435 * fixed bug 148: failure to inline-expand a local function left
2436 garbage, confusing the compiler.
2437 * fixed bugs 3cd: structure slot readers perform type check if the
2438 slot can have an invalid value (i.e. it is either not initialized
2439 or can be written with a less specific slot writer).
2440 * bug fix: the compiler now traps array references to elements off
2441 the end of an array; previously, the bounds checking in some
2442 circumstances could go off-by-one.
2443 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2445 * type declarations for array element types now obey the description
2446 on the CLHS page "Declaration TYPE", as per discussions on
2447 sbcl-help around 2003-05-08. This means that a declaration
2448 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2449 declaration, all references to BAR will be asserted or assumed
2450 (with THE, so dependent on compiler policy) to involve objects of
2451 type FOO. Note that no such declaration is implied in
2452 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2453 * declared types of functions from the "Conditions"
2454 chapter. (reported by Paul Dietz)
2455 * bug fix: CERROR accepts a function as its first argument.
2456 * bug fix: NTH an NTHCDR accept a bignum as index
2457 arguments. (reported by Adam Warner)
2458 * optimization: character compare routines now optimize comparing
2459 against a constant character. (reported by Gilbert Baumann)
2460 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2461 argument now works properly on the MIPS platform.
2462 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2463 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2464 declared to have a functional type, does not cause an error
2466 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2467 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2468 * fixed bug: if last continuation of a deleted block has a
2469 destination, this destination should be deleted too. (reported by
2471 * fixed a bug in the bootstrap process: the host compiler's values
2472 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2473 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2474 test case from Patrik Nordebo)
2475 * improved the ability of the disassembler on the PPC platform to
2476 provide helpful disassembly notes.
2477 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2478 the class in more cases than previously.
2479 * bug fix: FILE-POSITION now understands :START and :END for
2480 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2481 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2482 8) data. (thanks to David Lichteblau)
2483 * bug fix: it is possible to add a method to a generic function
2484 without lambda list.
2485 * bug fix: reader failed to signal END-OF-FILE inside an
2486 object representation. (reported by Nikodemus Siivola)
2487 * fixed some bugs revealed by Paul Dietz' test suite:
2488 ** LAST and [N]BUTLAST should accept a bignum.
2489 ** condition slot accessors are methods.
2490 ** (VECTOR NIL) is a subtype of STRING.
2492 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2493 * minor incompatible change: some nonsensical specialized lambda
2494 lists (used in DEFMETHOD) which were previously ignored now signal
2496 * minor incompatible change: the system is now aware of the types of
2497 variables in the COMMON-LISP package, and will signal errors for
2498 most violations of these type constraints (where previously they
2499 were silently accepted).
2500 * minor incompatible change: COMPILE-FILE now uses the freedom
2501 afforded (ANSI 3.2.2.3) to use derived function types for
2502 functions defined in the same file. This also permits the system
2503 to warn on static type mismatches and function
2504 redefinition. (Currently it does not work with high DEBUG level.)
2505 * minor incompatible change: VALUES declaration is disabled.
2506 * When issuing notes, the compiler now signals a condition of type
2507 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2508 restart for use in user handlers. It is expected that the
2509 COMPILER-NOTE condition will eventually become a condition
2510 supertype to a hierarchy of note types, which will then be
2511 handleable in a similar fashion. However, at the moment, no such
2512 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2513 but it's an implementation detail, not a classification for the
2515 * Changes in type checking closed the following bugs:
2516 ** type checking of unused values (192b, 194d, 203);
2517 ** template selection based on unsafe type assertions (192c, 236);
2518 ** type checking in branches (194bc).
2519 * A short form of VALUES type specifier has ANSI meaning (it has
2520 increased the number of situations when SBCL cannot perform type
2522 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2523 keywords or constants is permissible.
2524 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2525 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2526 classes. (thanks to Antonio Martinez)
2527 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2528 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2529 argument to a pretty-print stream. Instead, it leaves any such
2530 operations to DESCRIBE-OBJECT methods.
2531 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2533 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2534 lambda lists. (thanks to David Lichteblau)
2535 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2536 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2537 (thanks to Gerd Moellmann)
2538 * fixed bug 246: increased compilation speed of long
2539 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2541 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2543 * DEFCONSTANT now throws a condition of type
2544 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2545 constant to a non-EQL value; CONTINUE and ABORT restarts
2546 respectively change and preserve the value.
2547 * fixed bug 63: The code walker, part of the implementation of CLOS,
2548 is now better at handling symbol macros.
2549 * bug fix: There is no longer an internal implementation type named
2550 CL:LENGTH. (reported by Raymond Toy)
2551 * bug fix: In macro-like defining macros/special operators the
2552 implicit block does not enclose the lambda list.
2553 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2554 enclosed in parentheses) are not suitable as type specifiers, and
2555 their use properly signals an error now.
2556 * bug fix: An argument count mismatch for a type specifier in code
2557 being compiled no longer causes an unhandled error at compile
2558 time, but signals a compile-time warning.
2559 * fixed simple vector readable printing
2560 * bug fix: DESCRIBE takes more care over whether the class
2561 precedence list slot of a class is bound before accessing it.
2562 (reported by Markus Krummenacker)
2563 * bug fix: FORMATTER can successfully compile pretty-printer format
2564 strings which use variants of the ~* directive inside.
2565 * bug fix: SEARCH now applies its TEST predicate to the elements of
2566 the arguments in the correct order. (thanks to Wolfhard Buss)
2567 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2568 in type inference. (thanks to Robert E. Brown)
2569 * bug fix: Reading in symbols with an explicit package name of ""
2570 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2571 * fixed some bugs revealed by Paul Dietz' test suite:
2572 ** NIL is now allowed as a structure slot name.
2573 ** Arbitrary numbers, not just REALs, are allowed in certain
2574 circumstances in LOOP for-as-arithmetic clauses.
2575 ** Multiple class redefinitions before slot access no longer
2576 causes a type error.
2577 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2578 association between the name and a class.
2579 ** Generic functions with non-standard method-combination and over
2580 five methods all of which return constants no longer return NIL
2581 after the first few invocations. (thanks to Gerd Moellmann)
2582 ** CALL-NEXT-METHOD with no arguments now passes the original
2583 values of the arguments, even in the presence of assignment.
2584 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2586 ** DISASSEMBLE works with closures and funcallable instances.
2587 ** ADD-METHOD now returns the generic function, not the new method.
2588 ** FIND-METHOD signals an error if the lengths of the specializers
2589 is incompatible with the generic function, even if the ERRORP
2591 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2592 which its argument is a member.
2593 ** DEFCLASS only redefines the class named by its class-name
2594 argument if that name is the proper name of the class;
2595 otherwise, it creates a new class.
2596 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2597 of the UNBOUND-SLOT condition to the name of the slot.
2598 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2599 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2600 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2601 SLOT-MAKUNBOUND in the specified fashion.
2603 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2604 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2605 cross-compilation host. As a consequence, we can now bootstrap our
2606 way up to SBCL starting with a bare gcc toolchain and human-readable
2607 source code (first the source to CLISP, then the source to SBCL).
2608 * A contributed module containing a partial implementation of the
2609 simple-streams interface has been included. (thanks to Rudi
2611 * A contributed module implementing the RFC1321 Message Digest
2612 Algorithm, known as MD5, has been included.
2613 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2614 no longer has any effect, as the code controlled by this feature
2615 has been deleted. (As far as we know, no-one has ever built using
2616 this feature, and its semantics were confused in any case).
2617 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2618 work on conditions (as required by the ANSI specification),
2619 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2620 expected behaviour on conditions. Users should note, however,
2621 that such behaviour is not required by the ANSI specification,
2622 and so use of this behaviour may render their code unportable.
2623 * fixed some bugs revealed by Paul Dietz' test suite:
2624 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2626 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2627 ** SXHASH on condition objects no longer returns NIL.
2628 ** :ALLOCATION :CLASS slots are better treated; their values are
2629 updated on class redefinition, and initforms inherited from
2630 superclasses are applied.
2631 ** REMOVE-METHOD returns its generic function argument even when
2632 no method was removed.
2633 ** SHARED-INITIALIZE now initializes the values of the requested
2634 slots, including those with :ALLOCATION :CLASS.
2635 ** ALLOCATE-INSTANCE now works on structure classes defined via
2636 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2638 ** SLOT-EXISTS-P now works on conditions, as well as structures
2640 ** MAKE-LOAD-FORM now has the required methods on
2641 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2642 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2643 keyword, and now implements the SLOT-NAMES argument.
2644 ** methods with &OPTIONAL arguments no longer allow too many
2645 arguments to be passed in the call without error.
2646 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2647 option is consistent with the required arguments of the generic
2648 function lambda list.
2649 * bug fix: REQUIRE accepts a string designator. (Thanks to
2651 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2652 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2653 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2654 * bug fix: The compiler now checks for duplicated variables in macro
2656 * bug fix: SETQ on globals returns the correct value.
2657 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2658 not a COMPILER-ERROR (followed by some other strange error on
2659 choosing the CONTINUE restart).
2660 * bug fix: make.sh and friends are now more consistent in the way that
2661 they look for GNU "make".
2663 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2664 * experimental native threads support (on x86 Linux >=2.4 only).
2665 This is not compiled in by default: you need to add :SB-THREAD to
2666 the target features. See the "Beyond ANSI" chapter of the manual
2668 * fix for longstanding nonANSIism: The old distinction between
2669 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2670 The return value from CL:FIND-CLASS is now a CLOS class, and
2671 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2672 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2673 * An interface to the MetaObject Protocol, as described in Kiczales,
2674 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2675 MIT Press, 1991, is available from the SB-MOP package.
2676 * incompatible change: the SB-PCL package should now be considered
2677 a private implementation detail, and no longer a semi-private MOP
2679 * minor incompatible change: due to rearrangement for threads, the
2680 control stack and binding stack are now allocated at arbitrary
2681 addresses instead of being hardcoded per-port. Users affected by
2682 this probably have to be doing advanced things with shared
2683 libraries, and will know who they are.
2684 * minor incompatible change: Previously, all --eval forms used were
2685 processed with READ before any of them were processed with EVAL.
2686 Now each --eval form is processed with both READ and EVAL before
2687 the next --eval form is processed. (Thus package operations like
2688 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2689 work as the user might reasonably expect.)
2690 * minor incompatible change: *STANDARD-INPUT* is now only an
2691 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2693 * minor incompatible change: Y-OR-N-P is now character-oriented,
2694 not line oriented. Also, YES-OR-NO-P now works without errors.
2695 (thanks to Antonio Martinez)
2696 * sb-aclrepl module improvements: an integrated inspector, added
2697 repl features, and a bug fix to :trace command.
2698 * Known functions, which cannot be open coded by the backend, are
2699 considered to be able to check types of their arguments. (fixing
2700 a bug report by Nathan J. Froyd)
2701 * fixed a bug in computing method discriminating functions: It is
2702 now possible to define methods specialized on classes which have
2703 forward-referenced superclasses. (thanks to Gerd Moellmann)
2704 * fixed evaluation order in optional entries (reported by Gilbert
2706 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2707 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2708 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2709 required-by-AMOP NAME argument, as well as CLASS and
2710 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2711 * fixed bug 20: DEFMETHOD can define methods using names that are
2712 not the proper names of classes to designate class specializers.
2713 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2714 Unix file descriptors, instead of blowing up. (thanks to Antonio
2716 * Garbage collection refactoring: user-visible change is that a
2717 call to the GC function during WITHOUT-GCING will not do garbage
2718 collection until the end of the WITHOUT-GCING. If you were doing
2719 this you were probably losing anyway.
2720 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2721 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2722 TYPEP the latter but not the former.
2723 * The compiler issues a full WARNING for calls to undefined functions
2724 with names from the CL package.
2725 * MAP-INTO for a vector destination is open coded. (reported by
2726 Brian Downing on c.l.l)
2727 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2728 documentation string.
2729 * fixed some bugs revealed by Paul Dietz' test suite:
2730 ** COPY-ALIST now signals an error if its argument is a dotted
2732 ** Condition slots are now accessed more correctly in the presence
2733 of multiple initargs for a given slot.
2734 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2735 exclude restarts of the same name associated with a different
2737 ** DEFCLASS of forward-referenced classes with another
2738 forward-referenced class in the superclasses list no longer
2740 ** Condition slots are now initialized once each, not multiple
2741 times. (thanks to Gerd Moellmann)
2742 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2744 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2745 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2746 arguments contain duplicated elements.
2747 ** RESTART-CASE understands local macros.
2748 ** RESTART-CASE associates exactly its own restarts with a condition.
2749 ** ENDP in safe mode checks its argument to be of type LIST.
2750 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2751 in question is unbound.
2752 ** Optimized MAKE-INSTANCE functions no longer cause internal
2753 assertion failures in the presence of duplicate initargs.
2754 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2755 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2756 propagated even in the ctor optimized implementation of
2758 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2760 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2761 * a better implementation of SXHASH on (simple) bit vectors,
2762 measured both in execution speed and in distribution of results
2763 over the positive fixnums, has been installed. Likewise, a better
2764 implementation of EQUAL for simple bit vectors is now available.
2765 * fixed CEILING optimization for a divisor of form 2^k.
2766 * fixed bug 240 (emitting extra style warnings "using the lexical
2767 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2768 by Antonio Martinez)
2769 * fixed SXHASH, giving different results for NIL depending on type
2770 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2771 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2772 variable. (found by Rolf Wester)
2773 * SBCL does not ignore type declarations for special
2774 variables. (reported by rif on c.l.l 2003-03-05)
2775 * some bug fixes in contrib/sb-aclrepl/
2776 * fixed some bugs revealed by Paul Dietz' test suite:
2777 ** a bug in the CONS type specifier, whereby the CAR and CDR
2778 types got intertwined, has been fixed;
2779 ** the type system is now able to reason about the interaction
2780 between INTEGER and RATIO types more completely;
2781 ** APPEND, [N]REVERSE and NRECONC check that those their
2782 arguments, which must be proper lists, are really so;
2783 ** An array specialized to be unable to hold elements has been
2784 implemented, as required -- yes, really -- by ANSI;
2785 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2786 on malformed property lists;
2788 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2789 * incompatible packaging change: in line with Unix convention,
2790 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2791 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2792 * REQUIRE and PROVIDE are now optionally capable of doing something
2793 useful. See the documentation string for REQUIRE.
2794 * infrastructure for a managed SBCL contrib system: contributed
2795 modules in this release include:
2796 ** the ASDF system definition facility;
2797 ** an interface to the BSD Sockets API;
2798 ** an ACL-like convenience interface to the repl;
2799 (thanks to Kevin Rosenberg)
2800 ** an implementation of ROTATE-BYTE, with efficient implementation
2802 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2803 gives a closer approximation to the right answer than previously.
2804 (thanks to Raymond Toy)
2805 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2806 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2807 optional environment arguments, as required by ANSI.
2808 * fixed bugs in other functions taking environment objects, allowing
2809 calls with an explicit NIL environment argument to be compiled
2811 * fixed bug 228: primary return values from
2812 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2813 COMPILE or FUNCTION.
2814 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2815 :TYPE will no longer signal errors on innocuous objects.
2816 * fixed bug 231b: SETQ is better at respecting type declarations in
2817 the lexical environment.
2818 * fixed a bug in DEFCLASS: classes named by symbols with no or
2819 unprintable packages can now be defined.
2820 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2821 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2822 * fixed compiler failure related to checking types of functions.
2823 (reported by Robert E. Brown)
2824 * the compiler is now much more consistent in its error-checking
2825 treatment of bounding index arguments to sequence functions: in
2826 (SAFETY 3) code, errors will be signalled in almost all cases if
2827 invalid sequence bounding indices are passed to functions defined
2828 by ANSI to operate on sequences.
2829 * fixed a bug in the build procedure: documentation of SBCL-specific
2830 packages is now preserved and available in the final Lisp image.
2831 * lifted FDEFINITION lookup out of loops in the implementation of
2832 many list operations. (thanks to Robert E. Brown)
2833 * fixed a bug in the reader: the #n# reader macro now works for
2834 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2835 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2836 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2837 E. Brown and Tony Martinez respectively)
2838 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2839 SLOT-BOUNDP: in particular, it is now able to optimize them much
2840 better, and is now not vulnerable to having packages renamed.
2841 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2842 always called when a slot is not present in an instance. (thanks
2844 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2845 longer signal an error on unreading a character following EOF on
2846 the previous constituent stream. (thanks to Tony Martinez)
2847 * fixed some bugs revealed by Paul Dietz' test suite:
2848 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2849 not just nonnegative fixnums;
2850 ** the logical bit-array operators such as BIT-AND now accept an
2851 explicit NIL for their "opt-arg" argument (to indicate a
2852 freshly-consed result bit-array);
2853 ** ELT now signals an error on an invalid sequence index in safe
2855 ** the type system is now cleverer about negations of numeric
2856 types, and consequently understands the BIGNUM and RATIO types
2858 ** the type system is now cleverer about the interaction between
2859 INTEGER and RATIO types: while bugs still remain, many more
2860 cases are accurately computed;
2861 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2862 if it is in the last clause;
2863 ** CONSTANTLY now correctly returns a side-effect-free function in
2865 ** DECLARE is no longer treated as a special-operator; in
2866 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2867 * incremented fasl file version number due to the change in the
2868 DEFSTRUCT-SLOT-DESCRIPTION structure.
2870 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2871 * minor incompatible change: code processed by the "interpreter" or
2872 EVAL now has a compilation optimization policy of (DEBUG 2)
2873 (changed from (DEBUG 1)) to improve debuggability of interactive
2874 development, and to allow the use of the debug RETURN command in
2876 * an experimental implementation of the RETURN command for the
2877 debugger has been included. (thanks to Frederik Kuivinen)
2878 * fixed bug 62: constraints were not propagated into a loop.
2879 * fixed bug in embedded calls of SORT (reported and investigated by
2881 * fixed some bugs revealed by Paul Dietz' test suite:
2882 ** printing and reading of arrays with some dimensions having
2883 length 0 (thanks to Gerd Moellmann);
2884 ** BOA constructor with &AUX argument without a default value does
2885 not cause a type error;
2886 ** CONSTANTP now returns true for all self-evaluating objects.
2888 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2889 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2890 accessors that are related by inheritance, as specified in the
2891 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2892 Valtteri Vuorikoski)
2893 * The compiler is now able to inline functions that were defined in
2894 a complex lexical environment (e.g. inside a MACROLET).
2895 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2897 * Support for the upcoming FreeBSD-5.0 release has been included.
2898 (thanks to Dag-Erling Smorgrav)
2899 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2900 effect when it is not in a toplevel context.
2901 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2902 stand a better chance of being correct. (thanks to Gerd
2904 * fixed bug in COERCE, which now signals an error on coercing a
2905 rational to a bounded real type which excludes the expected
2907 * The compiler is now able to derive types more accurately from the
2908 COERCE and COMPILE functions.
2909 * fixed bug 223: functional binding is considered to be constant
2910 only for symbols in the CL package.
2911 * fixed bug 231: SETQ did not check the type of a variable being set
2912 (reported by Robert E. Brown)
2913 * A new optimization for MAKE-INSTANCE has been included, fixing
2914 various bugs (including relating to :ALLOCATION :CLASS slots and
2915 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2917 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2918 ** As required by ANSI, LOOP now disallows anonymous collection
2919 clauses such as COLLECT I in conjunction with aggregate boolean
2920 clauses such as THEREIS (= I 1);
2921 ** LOOP now signals an error when any variable is reused in the
2922 same loop (including the potentially useful construct analogous
2923 to WITH A = 1 WITH A = (1+ A);
2924 ** IT is only a special loop symbol within the first clause of a
2925 conditional loop clause;
2926 ** LOOP with a typed iteration variable over a hashtable now
2927 signals a type error iff it should.
2928 * fixed some other bugs revealed by Paul Dietz' test suite:
2929 ** FILE-STREAM now names the class previously known as FD-STREAM;
2930 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2931 argument) no longer signals an error;
2932 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2933 of the slot symbol, rather than using the current package
2934 ((:CONC-NAME "") continues to intern the slot's name in the
2936 * incremented fasl file version number, because of the incompatible
2937 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2938 of the new implementation of DEFINE-COMPILER-MACRO.
2940 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2941 * Support for building SBCL for MIPS platforms running in
2942 little-endian mode has now been checked in, and basic
2943 functionality on said platforms verified.
2944 * minor incompatible change: PCL now records the pathname of a file
2945 in which methods and the like are defined, rather than its
2947 * minor incompatible change: TRUENAME now considers the truename of
2948 a file naming a directory to be the pathname with :DIRECTORY
2949 component indicating that directory.
2950 * minor incompatible change: a NAMED clause in the extended form of
2951 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2952 reason for the previous behaviour is unclear.
2953 * more systematization and improvement of CLOS and MOP conformance
2954 in PCL (thanks to Gerd Moellman and Pierre Mai):
2955 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2957 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2958 primary methods with no specializers;
2959 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2961 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2962 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2963 FORWARD-REFERENCED-CLASSes; error reporting on
2964 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2966 ** SXHASH on CLOS instances now uses a slot internal to the
2967 instance to return different numbers on distinct instances,
2968 while preserving the same return value through invocations of
2970 ** DEFMETHOD signals errors when methods with longer incongruent
2971 lambda lists are added to generic functions;
2972 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2973 CLASS, as specified in AMOP;
2974 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2975 on the order returned by the primary method for classes of
2976 class STANDARD-CLASS;
2977 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2978 * fixed some bugs shown by Paul Dietz' test suite:
2979 ** DOLIST puts its body in TAGBODY;
2980 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2982 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2983 value producing form;
2984 ** if more variables are given to PROGV than values, extra
2985 variables are bound and made to have no value;
2986 ** NSUBSTITUTE on list arguments gets the right answer with
2988 ** ELT signals an error of type TYPE-ERROR when the index argument
2989 is not a valid sequence index;
2990 ** LOOP signals (at macroexpansion time) an error of type
2991 PROGRAM-ERROR when duplicate variable names are found;
2992 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2993 ** FUNCALL of special-operators now cause an error of type
2995 ** PSETQ now works as required in the presence of side-effecting
2996 symbol-macro places;
2997 ** NCONC accepts any object as its last argument;
2998 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3000 ** loop-for-as-package does not require a package to be explicitely
3002 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3004 * fixed bug 166: compiler preserves "there is a way to go"
3005 invariant when deleting code.
3006 * fixed bug 172: macro lambda lists with required arguments after
3007 &REST arguments now cause an error to be signalled. (thanks to
3009 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3010 bug. (thanks to Matthew Danish)
3011 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3013 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3014 arguments are lexically rebound. (thanks to Gerd Moellmann and
3016 * fixed bug 194: error messages are now more informative when there
3017 is no primary method applicable in a call to a generic
3018 function. (thanks to Gerd Moellmann)
3019 * fixed bug in command line argument checking (thanks to Julian
3021 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3022 (thanks to Matthew Danish)
3023 * incremented fasl file version number, because of the
3024 SXHASH-related changes in the layout of CLOS data structures
3026 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3027 * minor incompatible change: The runtime (the Unix executable named
3028 "sbcl") is now much pickier about the .core files it will load.
3029 Essentially it now requires .core files to descend from the same
3030 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3031 as the runtime does. (The intent is to prevent the crashes which
3032 can occur, and which can even be reported as mysterious failures,
3033 when people patch the sources or change the build parameters
3034 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3035 sbcl and .core files.)
3036 * fixed bug: VALUES-LIST is no longer optimized away.
3037 * fixed bug 142: The FFI conversion of C string values to Lisp
3038 string values no longer conses excessively. (thanks to Nathan
3039 Froyd porting Raymond Toy's fix to CMU CL)
3040 * began to systematize and improve MOP conformance in PCL (thanks to
3041 Nathan Froyd, Gerd Moellman and Pierre Mai):
3042 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3044 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3045 ** FINALIZE-INHERITANCE is now called on class finalization;
3046 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3047 argument precedence order.
3048 * fixed bug 202: The compiler no longer fails on functions whose
3049 derived types contradict their declared type.
3050 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3051 so it can be non-toplevel.
3052 * The fasl file version number has changed (because of the new
3053 implementation of DEFMACRO).
3054 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3055 safe code, that any length requirement by their type-specifier
3056 argument is valid. The exceptions to this are described in bug
3058 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3059 functions now check the types of their inputs as required by ANSI.
3060 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3061 introduced symbol is DECLAREd to be SPECIAL.
3062 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3063 * fixed bug: :COUNT argument to sequence functions may be negative
3064 * fixed bug: body of DO-SYMBOLS may contain declarations
3065 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3066 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3067 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3068 symbol macro only once
3069 * fixed printing of call frame when argument list is unavailable
3070 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3071 * compiler no longer signals WARNING on unknown keyword
3074 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3075 * A beta-quality port to the mips architecture running Linux,
3076 based on the old CMUCL backend, has been made. It has been tested
3077 on a big-endian kernel, and works sufficiently well to be able to
3078 rebuild itself; it has not been tested in little-endian mode.
3079 * fixed an inconsistency between gencgc.c and purify.c which made
3080 dumping/loading .core files unreliable
3081 * fixed bug 120a: The compiler now deals correctly with IFs where
3082 the consequent is the same as the alternative, instead of
3083 misderiving the return type. (thanks to Alexey Dejneka)
3084 * fixed bug 113: Logical pathnames are now dumpable (the logical
3085 host is resolved at load-time, throwing an error if it is not
3087 * fixed bug 174: FORMAT's error message is slightly clearer when a
3088 non-printing character is used in a format directive.
3089 * fixed several bugs in compiler checking of type declarations, i.e.
3090 violations of the Python "declarations are assertions" principle
3091 (thanks to Alexey Dejneka)
3092 * fixed several bugs in PCL's error checking (thanks to Gerd
3094 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3096 * fixed bug in compilation of functions as first class values
3097 (thanks to Antonio Martinez-Shotton)
3098 * The compiler's handling TYPE-ERRORs which it can prove will
3099 inevitably happen at runtime has been cleaned up and corrected
3100 in several ways. (thanks to Alexey Dejneka)
3101 * improved argument type checking for various basic arithmetic
3102 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3103 lavished on them in the past that they can be compiled in many
3104 ways in different special cases
3105 * fixed bug 181: compiler checks validity of user supplied type
3107 * cleaned up code flushing in optimization: Function calls which
3108 should signal errors for safety purposes (e.g. which ANSI says
3109 should signal errors when their arguments are of incorrect type)
3110 are no longer optimized away.
3111 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3112 * incremented fasl file version number, because changes in the
3113 implementation of sequence functions like COERCE caused
3114 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3115 in old inline expansions) to become undefined. (Actually these
3116 changes were later undone, so we might very well be binary
3117 compatible with 0.7.7 after all, but leaving the version number
3118 incremented seemed like the simplest and most conservative
3121 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3122 * An alpha-quality port to the parisc architecture running Linux,
3123 based on the old CMUCL backend, has been made. This, even more so
3124 than the other backends, should be considered still a work in
3125 progress; known problems include that the Linux kernel in 64-bit
3126 mode does not propagate the correct sigcontext structure to
3127 userspace, and consequently SBCL on a parisc64 kernel will not
3129 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3130 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3131 suppressed.) Also now that the compiler is looking at declarations
3132 in the environment, it checks optimization declarations as well,
3133 and suppresses inlining when (> DEBUG SPEED).
3134 * More fixes have been made to treatment of floating point exception
3135 treatment and other Unix signals. In particular, floating point
3136 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3137 * The detection and handling of control stack exhaustion (infinite
3138 or very deeply nested recursion) has changed. Stack exhaustion
3139 detection is now done by write-protecting pages at the OS level
3140 and applies at all optimization settings; when found, a
3141 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3142 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3143 be caught using IGNORE-ERRORS.
3144 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3145 correctly when transforming recursive LABELS functions to LETs.
3146 (thanks to Alexey Dejneka)
3147 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3148 that are names of constants or global variables.
3149 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3150 alien routines with docstrings.
3151 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3152 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3154 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3155 (thanks to Eric Marsden)
3156 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3157 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3158 * Bug fix: The SPARC backend can now compile functions involving
3159 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3160 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3161 to a functions expecting a PATHNAME-DESIGNATOR.
3162 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3163 lambda lists. (thanks to Alexey Dejneka)
3164 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3165 functions so that EXPORTing the name of the class doesn't cause
3166 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3167 to Antonio Martinez for reporting this)
3168 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3169 constant-folded without failing an assertion. (thanks to Einar
3170 Floystad Dorum for reporting this)
3171 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3172 types are now dealt with more correctly by the compiler.
3173 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3174 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3175 behaviour analogous to RENAME-FILE. This puts its behaviour more
3176 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3178 * The fasl file version number has changed again. (because of the
3179 bug fix involving the names of PCL MAKE-INSTANCE functions)
3181 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3182 * bug fix: Floating point exceptions are treated much more
3183 consistently on the x86/Linux and PPC/Linux platforms.
3184 * Array initialization with :INITIAL-ELEMENT is now much faster for
3185 cases when the compiler cannot open code the array creation, but
3186 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3187 array accesses have also seen a speed increase.
3188 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3189 specified. (This was at the root of some bad interactions between
3190 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3192 * bug fix: Internal error arguments for undefined functions are now
3193 computed correctly on the PPC/Linux platform.
3194 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3195 Raymond Toy's patch for CMU CL)
3196 * Support for the Solaris 9 operating environment has been included
3197 (thanks to Daniel Merritt)
3198 * A very ugly but hopefully complete draft of the missing FFI chapter
3199 of the manual has been created by reformatting the corresponding
3200 CMU CL manual chapter into (currently very ugly and incoherent)
3201 DocBook and bringing it up to date for SBCL behavior. Thus, the
3202 manual is now essentially complete, at least by my extreme
3203 once-and-only-once standards, whereby it's acceptable to refer to
3204 the doc strings of SB-EXT functions as the primary documentation.
3205 * The fasl file version number has changed again, due to cleanup of
3206 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3207 is no longer a static symbol.)
3209 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3210 * SBCL now builds with OpenMCL (version 0.12) as the
3211 cross-compilation host; also, more progress has been made toward
3212 bootstrapping under CLISP.
3213 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3215 * bug 158 fixed: The compiler can now deal with integer loop
3216 increments different from 1; fixing this turned out also to fix
3218 * bug 169 fixed: no more bogus warnings about using lexical bindings
3219 despite the presence of perfectly good SPECIAL declarations (thanks
3220 to David Lichteblau)
3221 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3222 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3223 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3225 * bug 180 fixed: Method combination specifications no longer ignore
3226 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3227 * bug fix: Structure type predicate functions now check their argument
3228 count as they should.
3229 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3230 correctly. (thanks to Pierre Mai)
3231 * minor incompatible change: The --noprogrammer option is deprecated
3232 in favor of the new --disable-debugger option, which is very similar.
3233 (The major difference is that it takes effect at a slightly different
3234 time at startup, causing handling of errors in --sysinit and
3235 --userinit files will be affected differently.) The
3236 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3237 been added to allow this functionality to be controlled from ordinary
3238 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3239 maintainers, who might want to run non-interactive scripts to
3240 build SBCL cores which will later be used interactively.)
3241 * minor incompatible change: The LOAD function no longer, when given
3242 a wild pathname to load, loads all files matching that pathname.
3243 Instead, an error of type FILE-ERROR is signalled.
3245 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3246 * bug 147 fixed: The compiler preserves its block link/count
3247 invariants more correctly now so that it doesn't crash. (thanks
3249 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3251 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3252 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3253 no-value-for-this-character as NIL (as specified by ANSI).
3254 * HOST-NAMESTRING on physical pathnames now returns a string that is
3255 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3256 (thanks to Christophe Rhodes)
3257 * The Alpha port handles icache flushing more correctly. (thanks to
3259 * More progress has been made toward bootstrapping under CLISP. (thanks
3260 to Christophe Rhodes)
3261 * The fasl file format has changed again, because dynamic loading
3262 on OpenBSD (which has non-ELF object files) motivated some cleanups
3263 in the way that foreign symbols are transformed and passed around.
3264 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3265 is no longer treated as whitespace by the reader, but instead as
3266 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3267 |A
\7fB|, instead of A as it used to.
3269 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3270 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3271 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3272 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3273 this as of 0.7.1.45, but was left out of the previous news section
3274 (thanks to Dan Barlow)
3275 * SBCL now runs on the Solaris operating system on SPARC architectures
3276 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3277 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3278 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3279 * SPARC backend cleanups, allowing builds of cores optimized for V8
3280 and V9 SPARCS, and also emission of code targeted to a particular
3281 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3283 * SBCL is closer to bootstrapping under CLISP, thanks to various
3284 fixes by Christophe Rhodes.
3285 * The fasl file format has changed again, to allow the compiler's
3286 INFO database to support symbol macros.
3287 * The user manual (in doc/) is formatted into HTML more nicely.
3288 (thanks to coreythomas)
3289 * The system is smarter about SUBTYPEP relationships, especially
3290 those involving NOT types (including types such as ATOM which are
3291 represented internally using NOT types). Thus SUBTYPEP is less
3292 likely to return (VALUES NIL NIL) in general, and in particular
3293 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3295 * The fasl file format has changed again, because the internal
3296 representation of types now includes a new slot to support the new
3297 SUBTYPEP-of-NOT-types logic.
3298 * (not a change in the main branch of SBCL, but a related prototype
3299 which can hopefully be merged into the main branch of SBCL in the
3300 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3301 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3302 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3303 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3304 dispatch macro character combinations. (thanks to Alexey Dejneka)
3305 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3306 that superficially look like logical namestrings correctly.
3307 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3309 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3310 default supplied-p) for &optional and &key arguments. (thanks to
3313 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3314 * incompatible change: The compiler is now less aggressive about
3315 tail call optimization, doing it only when (> SPACE DEBUG) or
3316 (> SPEED DEBUG). (This is an incompatible change because there are
3317 programs which relied on the old CMU-CL-style behavior to optimize
3318 away their unbounded recursion which will now die of stack overflow.)
3319 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3320 for non-GENCGC systems has been increased to 20M (since that
3321 seems much closer to the likely performance optimum for modern
3322 systems than the old 4M value was)
3323 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3324 and *DEBUG-PRINT-LEVEL*
3325 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3326 of CMU CL's support for SPARC, and various endianness and other
3327 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3328 * new syntactic sugar for the Unix command line: --load foo.bar is now
3329 an alternate notation for --eval '(load "foo.bar")'.
3331 ** The system now detects stack overflow and handles it gracefully,
3332 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3333 optimization settings. (This is a good thing in general, and
3334 its introduction in this version should be particularly timely
3335 for anyone whose code fails because of suppression of tail
3337 ** The system now hunts for the C variable "environ" in a more
3338 devious way, to avoid segfaults when the C library version differs
3339 between compile time and run time. (thanks to Christophe Rhodes)
3340 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3341 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3342 ** The compiler no longer issues bogus style warnings for undefined
3343 classes in the same source file as the DEFCLASSes which defined
3344 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3346 ** fixes in CONDITION class precedence list for undefined function
3347 errors (thanks to Alexei Dejneka)
3348 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3349 correctly. (thanks to Dan Barlow)
3350 ** portability fixes aiming at bootstrapping under CLISP (thanks
3351 to Dave McDonald and Christophe Rhodes)
3352 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3353 ** fixes in type translation and and type inference (thanks to
3355 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3356 ** various fixes in the new ports (thanks to Dan Barlow)
3357 * several changes related to debugging:
3358 ** suppression of tail recursion, as noted above
3359 ** stack overflow detection, as noted above
3360 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3361 is now the default. (For some time encapsulation has been more
3362 reliable than the breakpoint-based :ENCAPSULATE NIL
3363 implementation, at least on X86 systems; and I just noticed that
3364 encapsulation also seems closer to the spirit of the ANSI
3367 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3369 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3370 up properly again. (There was a packaging bug in 0.7.0 which
3371 left their definitions in SB-SYS::LOAD-FOREIGN and
3372 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3373 vital for most things which interface to C-level interfaces,
3374 like extensions working with sockets or databases or
3375 Perl-compatible regexes or whatever, and the need to fix
3376 this bug is the main reason that 0.7.1 was released so
3378 ** DEFGENERIC is now choosier about the methods it redefines, so that
3379 reLOADing a previously-LOADed file containing DEFGENERICs does
3380 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3381 cycle now works as it should. (thanks to Alexey Dejneka)
3382 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3383 fixed by Christophe Rhodes. (He actually submitted this patch
3384 months ago, and I delayed until after 0.7.0.)
3385 ** Bug 111 (internal compiler confusion about runtime checks on
3386 FUNCTION types) was fixed by Alexey Dejneka.
3387 * Some internal cleanups (getting rid of variables which aren't
3388 needed now that the byte interpreter is gone) caused the fasl
3389 file format number to change again.
3391 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3392 * major incompatible change: The default fasl file extension, i.e. the
3393 default extension for files produced by COMPILE-FILE, has changed
3394 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3396 ** There are many changes in the implementation of the compiler.
3397 SBCL is now essentially a compiler-only implementation of ANSI
3398 Common Lisp. EVAL still "interprets" a few special cases, but
3399 almost all the interesting cases are handled by creating
3400 a LAMBDA expression, calling COMPILE on it, then calling
3401 FUNCALL on the result.
3402 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3403 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3404 Since the code is newer, there might still be some new bugs
3405 (though not as many as before Martin Atzmueller's fixes:-). But
3406 the new code is substantially simpler and clearer, and hopefully
3407 any remaining bugs will be simpler, less fundamental, and more
3408 fixable then the bugs in the old code.
3409 ** The revised compiler is still a little unsteady on its feet.
3411 *** The debugging information it produces (particularly the names
3412 of FUNCTION objects) is sometimes much less useful than what
3413 the old compiler produced.
3414 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3415 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3416 within a MACROLET) has been temporarily weakened.
3417 ** There are new compiler optimizations for various functions:
3418 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3419 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3420 *** the math functions TRUNCATE, FLOOR, and CEILING
3421 *** the function-of-all-trades COERCE
3422 Mostly these should be transparent, but there's one
3423 potentially-annoying problem (bug 117): when the compiler
3424 inline-expands a function and does type analysis on the result,
3425 it can create control paths which have type mismatches, and
3426 when it can't prove that those control paths aren't taken,
3427 it will issue WARNINGs about the type mismatches. This is
3428 a particular problem in practice for the new sequence functions.
3429 It's not clear how this should be fixed, and for now, a
3430 workaround is given in the entry for 117 in the BUGS file.
3431 ** (Because of the interaction between the two previous items --
3432 occasional inlining problems and new inline expansions -- some
3433 of the new sequence function optimizations won't really kick in
3434 completely until debugging information, and then inlining, are
3435 straightened out in some future version.)
3436 * minor incompatible changes:
3437 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3438 DIRECTORY no longer implicitly promotes NIL slots of its
3439 pathname argument to :WILD. In particular, when you ask for the
3440 contents of a directory (which you used to be able to do without
3441 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3442 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3443 ** changes in behavior that ANSI explicitly defines to be
3444 implementation dependent:
3445 *** The new compiler-only implementation still conforms with ANSI,
3446 but acts a little different than before. Besides the obvious
3447 changes in performance tradeoffs (that the cost per form passed
3448 to EVAL has gone up, and the cost per form executed by EVAL
3449 has gone down), the behavior of the system changes a little
3450 because there are no longer any interpreted function objects.
3451 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3452 e.g. doing COMPILE on the output of interactive DEFUN is
3454 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3456 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3457 has changed from (:CL) to NIL.
3458 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3459 e.g. #\Newline and #\Tab, don't have names specified in the
3460 ANSI Common Lisp standard, is now based on their ASCII symbolic
3461 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3462 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3463 are no longer used for output.
3464 ** changes in internal implementation constants:
3465 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3466 4 million. (If your application spends a lot of time GCing and
3467 you have a lot of RAM, you might want to experiment with
3468 increasing it even more.)
3469 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3470 However, almost all old code should still continue to work without
3471 immediate update, as SB-C-CALL is now a (deprecated) nickname
3473 ** Old operator names in the style DEF-FOO are now deprecated in
3474 favor of new corresponding names DEFINE-FOO, for consistency with
3475 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3476 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3477 internal symbols, but a few supported extensions like
3478 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3479 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3480 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3481 etc. as you get deeper into recursive calls to the debugger
3482 command loop, instead of the old "5]", "5]]", "5]]]"
3483 sequence. (I was motivated to do this when squabbles between
3484 ILISP and SBCL left me very deeply nested in the debugger. In the
3485 short term, this change will probably provoke more ILISP/SBCL
3486 squabbles, but hopefully it will be an improvement in the long run.)
3487 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3488 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3489 function, since declaiming return types involving aliens is
3490 (1) annoyingly messy to do by hand and (2) vital to efficient
3491 compilation of code which calls such functions.
3492 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3493 longer reexported by the SB-EXT package. They're solely useful
3494 for alien code, so it seems more logical that you should get
3495 them from the SB-ALIEN package, not in SB-EXT.
3496 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3497 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3498 features. Instead, the code that they used to control is always
3499 built into the system.
3500 * many other bug fixes
3501 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3502 advantage of the new EVAL-WHEN stuff and to clean them up in
3503 general, and they are now more ANSI-compliant in a number of
3504 ways. Martin Atzmueller is responsible for a lot of this.
3505 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3507 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3508 *** correct ERROR type for various file operations
3509 *** some fixes for Lisp streams
3510 *** DEFMETHOD syntax checking
3511 *** changing old weird representation of debug information as
3512 strings (which, among their other deficiencies, don't transform
3513 correctly when you rename packages, and don't change their
3514 print representation when you change things like *PACKAGE*
3515 and *PRINT-LENGTH*) to symbols and lists of symbols
3516 He also made several improvements and fixed several bugs in DESCRIBE.
3517 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3519 *** misbehavior of WRITE-STRING/WRITE-LINE
3520 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3521 and several other LOOP problems as well
3522 *** DIRECTORY when similar filenames are present
3523 *** DEFGENERIC with :METHOD options
3524 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3525 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3526 *** argument ordering in FIND with :TEST option
3527 *** mishandled package designator argument in APROPOS-LIST
3528 *** various problems in the backquote readmacro
3529 *** a bug in APROPOS
3530 *** probably some others that I'm not describing very well here,
3531 since the CVS log documents them by reference to sbcl-devel
3532 messages, and the SourceForge archives aren't working well.:-(
3533 ** Dan Barlow improved the Alpha port (and is making progress on the
3534 PPC port, for those of you who think different).
3535 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3536 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3537 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3538 continues to work on the SPARC port (for those of you in a position
3539 to look down upon our little PC-compatible boxes from a great height).
3540 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3541 and uses that copy, rather than the current dynamic value, when
3542 it's trying to decide whether to truncate output. Thus e.g.
3543 (let ((*print-lines* 50))
3544 (pprint-logical-block (stream nil)
3546 (let ((*print-lines* 8))
3547 (print (aref possiblybigthings i) stream)))))
3548 should now truncate the logical block only at 50 lines, instead of
3549 often truncating it at 8 lines, as it did before.
3550 * The doc/cmucl/ directory, containing old CMU CL documentation
3551 from the time of the fork, is no longer part of the base system.
3552 SourceForge has shut down its anonymous FTP service, and with it
3553 my original plan for distributing the old CMU CL documentation
3554 there. For now, if you need these files you can download an old
3555 SBCL source release and extract them from it.
3556 * The fasl file version number changed again, for dozens of reasons,
3557 some of which are apparent above.
3559 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3560 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3561 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3562 patch, so that DISASSEMBLE output is much nicer.
3563 * The code in the SB-PROFILE package now seems reasonably stable.
3564 I still haven't decided what the final interface should look like
3565 (I'd like PROFILE to interact cleanly with TRACE, since both
3566 facilities use function encapsulation) but if you have a need
3567 for profiling now, you can probably use it successfully with
3568 the current CMU-CL-style interface.
3569 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3570 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3571 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3572 link now returns the dangling link itself, and for similar
3573 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3574 link itself. (In these cases the old code signalled an error and
3575 looped endlessly, respectively.) Thus, DIRECTORY now works even
3576 in the presence of dangling and cyclic symbolic links.
3577 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3578 is now a supported extension again, since the consensus on
3579 sbcl-devel was that it can be useful for ordinary development
3580 work, not just for debugging SBCL itself.
3581 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3582 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3583 that currently-defined functions might be redefined later with
3584 different return types.
3585 * Hash tables can be printed readably, as inspired by CMU CL code
3586 of Eric Marsden and SBCL code of Martin Atzmueller.
3587 * better error handling in CLOS method combination, thanks to
3588 Martin Atzmueller porting Pierre Mai's CMU CL patches
3589 * more overflow fixes for >16Mbyte I/O buffers
3590 * A bug in READ has been fixed, so that now a single Ctrl-D
3591 character suffices to cause end-of-file on character streams.
3592 In particular, now you only need one Ctrl-D at the command
3593 line (not two) to exit SBCL.
3594 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3596 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3597 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3598 does the right thing.
3599 * The implementation of some type tests, especially for CONDITION
3600 types, is now tidier and maybe faster, due to CMU CL code
3601 originally by Douglas Crosher, ported by Martin Atzmueller.
3602 * Some math functions have been fixed, and there are new
3603 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3604 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3605 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3606 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3607 likely to be built into sbcl-0.7.0.)
3608 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3609 CL, ported to SBCL by Martin Atzmueller)
3610 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3611 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3612 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3613 * There's a new slam.sh hack to shorten the edit/compile/debug
3614 cycle for low-level changes to SBCL itself, and a new
3615 :SB-AFTER-XC-CORE target feature to control the generation of
3616 the after-xc.core file needed by slam.sh.
3617 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3618 COMPILE-FILE is no longer supported, so that now every function
3619 gets an entry point, so that block compilation looks a little
3620 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3621 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3622 returns the number of bytes consed since the system started,
3623 rather than the number consed since the first time the function
3624 was called. (The new definition parallels ANSI functions like
3625 CL:GET-INTERNAL-RUN-TIME.)
3626 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3627 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3628 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3629 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3630 * Fasl file version numbers are now independent of the target CPU,
3631 since historically most system changes which required version
3632 number changes have affected all CPUs equally. Similarly,
3633 the byte fasl file version is now equal to the ordinary
3636 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3637 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3638 is no longer recognized. I apologize for this, because it was
3639 listed in SB-EXT as a supported extension, but I found that
3640 its existing behavior was poorly specified, as well as incorrectly
3641 specified, and it looked like too much of a mess to straighten it
3642 out. I have enough on my hands trying to get ANSI stuff to work..
3643 * many patches ported from CMU CL by Martin Atzmueller, with
3644 half a dozen bug fixes in pretty-printing and the debugger, and
3645 half a dozen others elsewhere
3646 * fixed bug 13: Floating point infinities are now supported again.
3647 They might still be a little bit flaky, but thanks to bug reports
3648 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3649 as flaky as they were.
3650 * The --noprogrammer command line option is now supported. (Its
3651 behavior is slightly different in detail from what the old man
3652 page claimed it would do, but it's still appropriate under the
3653 same circumstances that the man page talks about.)
3654 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3655 are now supported, and enabled by default. Thus, the compiler can
3656 handle many floating point and complex operations much less
3657 inefficiently. (Thus e.g. you can implement a complex FFT
3659 * The compiler now detects type mismatches between DECLAIM FTYPE
3660 and DEFUN better, and implements CHECK-TYPE more correctly, and
3661 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3662 thanks to patches from Martin Atzmueller.
3663 * various fixes to make the cross-compiler more portable to
3664 ANSI-conforming-but-different cross-compilation hosts (notably
3665 Lispworks for Windows, following bug reports from Arthur Lemmens)
3666 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3667 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3668 fixed thanks to Pierre Mai's CMU CL patches.
3669 * fixes to keep the system from overflowing internal counters when
3670 it tries to use i/o buffers larger than 16M bytes
3671 * fixed bug 45a: Various internal functions required to support
3672 complex special functions have been merged from CMU CL sources.
3673 (When I was first setting up SBCL, I misunderstood a compile-time
3674 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3675 * improved support for type intersection and union, fixing bug 12
3676 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3677 more obscure bugs as well
3678 * some steps toward byte-compiling non-performance-critical
3679 parts of the system, courtesy of patches from Martin Atzmueller
3680 * Christophe Rhodes has made some debian packages of sbcl at
3681 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3682 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3683 stable, but are nonetheless usable. When he's ready, I'd be happy
3684 to add them to the SourceForge "File Releases" section. (And if
3685 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3686 * new fasl file format version number (because of changes in
3687 internal representation of (OR ..) types to accommodate the new
3688 support for (AND ..) types, among other things)
3690 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3691 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3693 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3695 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3696 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3697 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3698 * DESCRIBE now works on structure objects again.
3699 * Most function call argument type mismatches are now handled as
3700 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3701 know whether the function will be redefined before the call is
3702 executed. (The compiler could flag local calls with full WARNINGs,
3703 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3704 it doesn't keep track of enough information to know whether calls
3705 are local in this sense.)
3706 * Compiler output is now more verbose, with messages truncated
3707 later than before. (There should be some supported way for users
3708 to override the default verbosity, but I haven't decided how to
3709 provide it yet, so this behavior is still controlled by the internal
3710 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3711 src/compiler/ir1util.lisp.)
3712 * Fasl file format version numbers have increased again, because
3713 support for the Gray streams extension changes the layout of the
3714 system's STREAM objects.
3715 * The Gray subclassable streams extension now works, thanks to a
3716 patch from Martin Atzmueller.
3717 * The full LOAD-FOREIGN extension (not just the primitive
3718 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3719 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3720 but like most other programs, it defaults to copying the Unix
3721 environment from the original process instead of starting the
3722 new process in an empty environment.
3723 * Extensions which manipulate the Unix environment now support
3724 an :ENVIRONMENT keyword option which doesn't smash case or
3725 do other bad things. The CMU-CL-style :ENV option is retained
3726 for porting convenience.
3727 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3728 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3730 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3732 * A patch from Martin Atzmueller seems to have solved the SIGINT
3733 problem, and as far as we know, signal-handling now works cleanly.
3734 (If you find any new bugs, please report them!)
3735 * The system no longer defaults Lisp source file names to types
3736 ".l", ".cl", or ".lsp", but only to ".lisp".
3737 * The compiler no longer uses special default file extensions for
3738 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3739 seems to expect a single default extension for all compiled code,
3740 and there's no compelling reason to try to stretch the standard
3741 to allow two different extensions.) Instead, byte-compiled files
3742 default to the same extension as native-compiled files.
3743 * Fasl file format version numbers have increased again, because
3744 a rearrangement of internal implementation packages made some
3745 dumped symbols in old fasl files unreadable in new cores.
3746 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3747 many fewer weird special cases.
3748 * Bug #17 (differing COMPILE-FILE behavior between logical and
3749 physical pathnames) has been fixed, and some related misbehavior too,
3750 thanks to a patch from Martin Atzmueller.
3751 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3752 by Tim Moore, ported to SBCL by Martin Atzmueller.
3753 * Martin Atzmueller fixed several filesystem-related problems,
3754 including bug #36, in part by porting CMU CL patches, which were
3755 written in part by Paul Werkowski.
3756 * More compiler warnings in src/runtime/ are gone, thanks to
3757 more patches from Martin Atzmueller.
3758 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3761 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3763 * DESCRIBE now works on CONDITION objects.
3764 * The debugger now handles errors which arise when trying to print
3765 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3767 * The build system now uses an additional file, customize-target-features.lisp,
3768 to allow local modifications to the target *FEATURES* list. (The point of
3769 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3770 debugging features enabled, without having to worry about propagating it
3771 into everyone's system when I do a "cvs update".) When no
3772 customize-target-features.lisp file exists, the target *FEATURES* list
3773 should be constructed the same way as before.
3774 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3775 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3776 warn about behavior which is undefined under the ANSI spec. Note: This
3777 is specified by ANSI, but it's not very popular with programmers.
3778 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3779 macro in the SBCL sources for an example of a workaround which you
3780 might use to make portable ANSI-standard code which does what you want.
3781 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3782 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3783 when it's used as a non-toplevel form. (This is required in order
3784 to implement the DEFCONSTANT-EQX macro.)
3785 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3786 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3787 funny things, due to bugs in the implementation of EVAL-WHEN
3788 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3789 non-toplevel DEFCONSTANTs are uncommon.)
3790 * The core file version number and fasl file version number have been
3791 incremented, because the old noncompliant DEFCONSTANT behavior involved
3792 calling functions which no longer exist, and because I also took the
3793 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3794 * fixed bug 1 (error handling before read-eval-print loop starts), and
3795 redid debugger restarts and related debugger commands somewhat while
3797 ** The QUIT debugger command is gone, since it did something
3798 rather different than the SB-EXT:QUIT command, and since it never
3799 worked properly outside the main toplevel read/eval/print loop.
3800 Invoking the new TOPLEVEL restart provides the same functionality.
3801 ** The GO debugger command is also gone, since you can just invoke
3802 the CONTINUE restart directly instead.
3803 ** The TOP debugger command is also gone, since it's redundant with the
3804 FRAME 0 command, and since it interfered with abbreviations for the
3806 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3808 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3809 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3810 some porting work by Martin Atzmueller)
3811 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3812 with a boatload of patches which he ported to SBCL. Now that those
3814 ** The system tries to make sure that its low-priority messages
3815 are prefixed by semicolons, to help people who like to use
3816 syntax highlighting in their ILISP buffer. (This patch
3817 was originally due to Raymond Toy.)
3818 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3819 patches originally written by Raymond Toy.
3820 ** The compiler understands coercion between single-value and
3821 multiple-VALUES type expressions better, getting rid of some very
3822 weird behavior, thanks to patches originally by Robert MacLachlan
3823 and Douglas Crosher.
3824 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3825 lambda lists, thanks to a patch originally by Pierre Mai.
3826 ** The system no longer bogusly warns about "abbreviated type
3828 ** The compiler gets less confused by inlining and RETURN-FROM,
3829 thanks to some patches originally by Tim Moore.
3830 ** The system no longer hangs when dumping circular lists to fasl
3831 files, thanks to a patch originally from Douglas Crosher.
3832 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3833 undefined function error.
3834 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3835 in structure slot initforms) for now, documented workaround instead:-|
3836 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3837 * fixed bug 5: added stubs for various Gray stream functions called
3838 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3839 installed, at least appropriate type errors are generated
3840 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3841 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3842 consistently in DEFMETHOD forms.
3843 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3844 it doesn't seem to affect SBCL after all
3845 * The C runtime system now builds with better optimization and many
3846 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3848 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3850 * The system is now under CVS at SourceForge (instead of the
3851 CVS repository on my home machine).
3852 * The new signal handling code has been tweaked to treat register
3853 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3854 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3855 downstream, e.g. in debug-int.lisp, has implicit dependencies
3856 on the unsignedness of integer representation of machine words,
3857 and that caused the system to bomb out with infinite regress
3858 when trying to recover from type errors involving signed values,
3859 e.g. (BUTLAST '(1 2 3) -1).)
3860 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3861 (This was one of the bugs Peter Van Eynde reported back in July.)
3862 * The system now uses code inspired by Colin Walters' O(N)
3863 implementation of MAP (from the cmucl-imp@cons.org mailing
3864 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3865 inline the MAP operation, and there is more than one
3866 sequence argument to the MAP call (so that it can't just
3867 do ETYPECASE once and for all based on the type of the
3868 single sequence argument). (The old non-inline implementation
3869 of the general M-argument sequence-of-length-N case required
3870 O(M*N*N) time when any of the sequence arguments were LISTs.)
3871 * The QUIT :UNIX-CODE keyword argument has been renamed to
3872 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3873 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3875 * Raymond Wiker's patches to port dynamic loading from Linux to
3876 FreeBSD have been added.
3877 * The BUGS file is now more nearly up to date, thanks in large part
3878 to Martin Atzmueller's review of it.
3879 * The debugger now flushes standard output streams before it begins
3880 its output ("debugger invoked" and so forth).
3881 * The core version number and fasl file version number have both
3882 been incremented, because of incompatible changes in the layout
3884 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3885 used to not be called for a saved Lisp image.)
3886 * Martin Atzmueller's version of a patch to fix a compiler crash,
3887 as posted on sbcl-devel 13 September 2000, has been installed.
3888 * Instead of installing Martin Atzmueller's patch for the
3889 compiler transform for SUBSEQ, I deleted the compiler transform,
3890 and transforms for some similar consing operations.
3891 * A bug in signal handling which kept TRACE from working on OpenBSD
3893 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3894 away by constant folding
3895 * The system now defines its address space constants in one place
3896 (in the Lisp sources), and propagates them automatically elsewhere
3897 (through GENESIS and the sbcl.h file). Therefore, patching the
3898 address map is less unnecessarily tedious and error-prone. The
3899 Lisp names of address space constants have also been systematized.
3900 * CVS tags like dollar-Header-dollar have been removed from
3901 the sources, because they have never saved me trouble and
3902 they've been source of trouble working with patches and other
3903 diff-related operations.
3904 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3905 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3907 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3909 * The system has been ported to OpenBSD.
3910 * The system now compiles with a simple "sh make.sh" on the systems
3911 that it's supported on. I.e., now you no longer need to tweak
3912 text in the target-features.lisp-expr and symlinks in src/runtime/
3913 by hand, the make.sh takes care of it for you.
3914 * The system is no longer so grossly inefficient when compiling code
3915 involving vectors implemented as general (not simple) vectors (VECTOR T),
3916 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3917 which dares to use the various sequence functions on non-simple
3918 vectors, takes less of a performance hit.
3919 * There is now a primitive type predicate VECTOR-T-P
3920 to test for the (VECTOR T) type, so that e.g.
3921 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3922 can now be compiled with some semblance of efficiency. (The old code
3923 turned the type declaration into a full call to %TYPEP at runtime!)
3924 * AREF on (VECTOR T) is still not fast, since it's still compiled
3925 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3926 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3927 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3928 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3929 without making full calls to SUBTYPEP at runtime.
3930 (Some analogous efficiency issues for non-simple vectors specialized to
3931 element types other than T, or for non-simple multidimensional arrays,
3932 have not been addressed. They could almost certainly be handled the
3933 same way if anyone is motivated to do so.)
3934 * The changes in array handling break binary compatibility, so
3935 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3936 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3937 instead of (VALUES T T).
3938 * By following the instructions that Dan Barlow posted to sbcl-devel
3939 on 2 July 2000, I was able to enable primitive dynamic object
3940 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3941 functionality is not implemented (since it calls ld to resolve
3942 library references automatically, requiring RUN-PROGRAM for its
3943 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3944 try to resolve library references) is now supported.
3945 * The system now flushes the standard output streams when it terminates,
3946 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3947 them at several other probably-convenient times, e.g. in each pass of
3948 the toplevel read-eval-print loop, and after evaluating a form given
3949 as an "--eval" command-line option. (These changes were motivated by a
3950 discussion of stream flushing issues on cmucl-imp in August 2000.)
3951 * The source transform for TYPEP of array types no longer assumes
3952 that an array whose element type is a not-yet-defined type
3953 is implemented as an array of T, but instead punts, so that the
3954 type will be interpreted at runtime.
3955 * There is now some support for cross-compiling in make.sh: each of
3956 the phases of make.sh has its own script. (This should be transparent
3957 to people doing ordinary, non-cross-compile builds.)
3958 * Since my laptop doesn't have hundreds of megabytes of memory like
3959 my desktop machine, I became more motivated to do some items on
3960 my to-do list in order to reduce the size of the system a little:
3961 ** Arrange for various needed-only-at-cold-init things to be
3962 uninterned after cold init. To support this, those things have
3963 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3964 symbols with such names are now uninterned after cold init).
3965 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3966 fasl files for cold load.
3967 ** Remove the old compiler structure pooling code (which used to
3968 be conditional on the target feature :SB-ALLOC) completely.
3969 ** Redo the representation of some data in cold init to be more compact.
3970 (I also looked into supporting byte compiled code at bootstrap time,
3971 which would probably reduce the size of the system a lot, but that
3972 looked too complicated, so I punted for now.)
3973 * The maximum signal nesting depth in the src/runtime/ support code has
3974 been reduced from 4096 to 256. (I don't know any reason for the very
3975 large old value. If the new smaller value turns out to break something,
3976 I'll probably just bump it back up.)
3977 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3979 * Many, many bugs reported by Peter Van Eynde have been added to
3980 the BUGS list; some have even been fixed.
3981 * While enabling dynamic object file loading, I tried to make the
3982 code easier to understand, renaming various functions and variables
3983 with less ambiguous names, and changing some function calling
3984 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3985 * While trying to figure out how to do the OpenBSD port, I tried to
3986 clean up some of the code in src/runtime/. In particular, I dropped
3987 support for non-POSIX signal handling, added various comments,
3988 tweaked the code to reduce the number of compilation warnings, and
3989 renamed some files to increase consistency.
3990 * To support the new automatic configuration functionality in make.sh,
3991 the source file target-features.lisp-expr has been replaced with the
3992 source file base-target-features.lisp-expr and the machine-generated
3993 file local-target-features.lisp-expr.
3994 * fixed a stupid quoting error in make.sh so that using CMU CL
3995 "lisp -batch" as cross-compilation host works again
3997 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3999 * DESCRIBE no longer tries to call itself recursively to describe
4000 bound/fbound values, so that it no longer fails on symbols which are
4001 bound to themselves (like keywords, T, and NIL).
4002 * DESCRIBE now works on generic functions.
4003 * The printer now prints less-screwed-up representations of closures
4004 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4005 * A private symbol is used instead of the :EMPTY keyword previously
4006 used to mark empty slots in hash tables. Thus
4007 (DEFVAR *HT* (MAKE-HASH-TABLE))
4008 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4009 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4010 now does what ANSI says that it should. (You can still get
4011 similar noncompliant behavior if bang on the hash table
4012 implementation with all the symbols you get back from
4013 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4014 This breaks binary compatibility, since tests for equality to
4015 :EMPTY are wired into things like the macroexpansion of
4016 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4018 * There's now a minimal placeholder implementation for CL:STEP,
4019 as required by ANSI.
4020 * An obscure bug in the interaction of the normal compiler, the byte
4021 compiler, inlining, and structure predicates has been patched
4022 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4023 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4025 * Missing ordinary arguments in a macro call are now detected even
4026 when the macro lambda list contains &KEY or &REST.
4027 * The debugger no longer complains about encountering the top of the
4028 stack when you type "FRAME 0" to explicitly instruct it to go to
4029 the top of the stack. And it now prints the frame you request even
4030 if it's the current frame (instead of saying "You are here.").
4031 * As specified by ANSI, the system now always prints keywords
4032 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4034 * The default initial SIZE of HASH-TABLEs is now smaller.
4035 * Type information from CLOS class dispatch is now propagated
4036 into DEFMETHOD bodies, so that e.g.
4037 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4039 is now basically equivalent to
4040 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4041 (DECLARE (TYPE SINGLE-FLOAT X))
4043 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4044 operation, without having to do run-time type dispatch.
4045 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4046 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4047 * Since I don't seem to be making big file reorganizations very often
4048 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4049 is overflowing my ability to conveniently back them up), I've finally
4050 checked the system into CVS. (The CVS repository is on my home system,
4051 not at SourceForge -- putting it on SourceForge might come later.)
4052 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4053 high-level GC-NOTIFY-FOO functions send their output. (There's
4054 still very little control of where low-level verbose GC functions
4055 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4056 less than it used to -- the GC-NOTIFY-FOO functions are now under
4057 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4058 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4059 in only one place in the source code, and propagates it automatically
4060 everywhere that it's needed. Thus e.g. when I bump the version from
4061 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4062 * The C source files now include boilerplate legalese and documentation
4063 at the head of each file (just as the Lisp source files already did).
4064 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4065 to his page will be replaced with a link to his new CLiki service.
4067 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4069 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4070 * The build process now looks for GNU make under the default name "gmake",
4071 instead of "make" as it used to. If GNU make is not available as "gmake"
4072 on your system, you can change this default behavior by setting the
4073 GNUMAKE environment variable.
4074 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4075 can build without error under CMU CL.
4077 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4079 * There is now a partial SBCL user manual (with some new text and some
4080 text cribbed from the CMU CL manual).
4081 * The beginnings of a profiler have been added (starting with the
4082 CMU CL profiler and simplifying and cleaning up). Eventually the
4083 main interface should be through the TRACE macro, but for now,
4084 it's still accessed through vaguely CMU-CL-style functions and macros
4085 exported from the package SB-PROFILE.
4086 * Some problems left over from porting CMU CL to the new
4087 cross-compilation bootstrap process have been cleaned up:
4088 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4089 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4090 part when getting rid of the compiler *BACKEND* stuff.)
4091 ** The value of *NULL-TYPE* was screwed up, because it was
4092 being initialized before the type system knew the final
4093 definition of the 'NULL type. This screwed up several key
4094 optimizations in the compiler, causing inefficiency in all sorts
4095 of places. (I found it because I wanted to understand why
4096 GET-INTERNAL-RUN-TIME was consing.)
4097 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4098 PROCLAIM FTYPE information. Unfortunately this broke binary
4099 compatibility again, since now the forms output by DEFGENERIC
4100 to refer to functions which didn't exist in 0.6.3.
4101 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4102 can use the new (as of 0.6.3) transform for SOME into MAP into
4104 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4105 values to (INTEGER 0 1000000), so that the time code will no longer
4106 occasionally get blown up by Linux returning 1000000 microseconds
4107 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4108 its output conform to the ANSI spec. (Alas, this makes its output
4109 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4110 * A full call to MAP NIL with a single sequence argument no longer conses.
4111 * fixes to problems pointed out by Martin Atzmueller:
4112 * The manual page no longer talks about multiprocessing as though
4113 it were currently supported.
4114 * The ILISP support patches have been removed from the distribution,
4115 because as of version 5.10.1, ILISP now supports SBCL without us
4116 having to maintain patches.
4117 * added a modified version of Raymond Toy's recent CMU CL patch for
4118 EQUALP comparison of HASH-TABLE
4120 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4122 * The system still can't cross-compile itself with
4123 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4124 entails), but at least it can compile more of itself that way
4125 than it used to be able to, and various buglets which were uncovered
4126 by trying to cross-compile itself that way have now been fixed.
4127 * This release breaks binary compatibility again. This time
4128 at least I've incremented the FASL file format version to 2, so that the
4129 problem can be detected reliably instead of just causing weird errors.
4130 * various new style warnings:
4131 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4132 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4133 ** using the lexical binding of a variable named in the *FOO* style
4134 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4135 as specified by ANSI.
4136 * *RANDOM-STATE* is no longer automatically initialized from
4137 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4138 default behavior of the system is to repeat its behavior every time
4139 it's run. If you'd like to change this behavior, you can always
4140 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4141 old convention there was no comparably easy way to get the system to
4142 repeat its behavior every time it was run.
4143 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4144 FTYPE declarations has been removed, in favor of their ANSI
4145 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4146 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4147 the types of their sequence arguments can be determined at compile time.
4148 This is done through a new open code expansion for MAP which eliminates
4149 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4150 argument types can be determined at compile time.
4151 * The optimizer now transforms COERCE into an identity operation when it
4152 can prove that the coerced object is already of the correct type. (This
4153 can be a win for machine generated code, including the output of other
4154 optimization transforms, such as the MAP transform above.)
4155 * Credit information has been moved from source file headers into CREDITS.
4156 * Source file headers have been made more standard.
4157 * The CASE macro now compiles without complaining even when it has
4160 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4162 * (Note that the way that the PCL macroexpansions were rewritten
4163 to accommodate the change in DEFGENERIC below breaks binary
4164 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4165 not run under sbcl-0.6.2. Once we get out of alpha releases,
4166 i.e. hit release 1.0.0, we'll probably try to maintain binary
4167 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4168 and sbcl-1.4.4. Until then, however, it might be fairly common
4169 for maintenance releases to break binary compatibility.)
4170 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4172 * The reporting of unhandled signals has been changed to print some
4173 explanatory text as well as the report form. (Previously only
4174 the report form was printed.)
4175 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4176 it defines, so that the compiler no longer issues undefined function
4177 warnings for compiled-but-not-yet-loaded generic functions.
4178 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4179 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4180 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4181 section of the ANSI spec are supported.
4182 * The "" nickname for the "KEYWORD" package has been removed.
4183 The reader still handles symbol tokens which begin with a package marker
4184 as keywords, but it doesn't expose its mechanism for doing so in the
4185 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4186 * The system now issues STYLE-WARNINGs for contradictory TYPE
4187 proclamations. (Warnings for contradictory FTYPE proclamations would
4188 be nice too, but those can't be done usefully unless the type system
4189 is made smarter about FUNCTION types.)
4190 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4191 systematized, so that "*target-*.lisp is supposed to exist only on the
4192 target and imply that there's a related file which exists on the
4193 host, and *host-*.lisp is supposed to exist only on the host and imply
4194 that there's a related file which exists on the target. This involves a
4195 lot of renaming. Hopefully the acute confusion caused by the renaming
4196 will be justified by the reduction in chronic confusion..
4197 ** runtime-type.lisp -> early-target-type.lisp
4198 ** target-type.lisp -> late-target-type.lisp
4199 ** early-host-format.lisp -> early-format.lisp
4200 ** late-host-format.lisp -> late-format.lisp
4201 ** host-error.lisp -> misc-error.lisp
4202 ** early-error.lisp -> early-target-error.lisp
4203 ** late-error.lisp -> late-target-error.lisp
4204 ** host-defboot.lisp -> early-defboot.lisp
4205 ** code/misc.lisp -> code/target-misc.lisp
4206 ** code/host-misc.lisp -> code/misc.lisp
4207 ** code/numbers.lisp -> code/target-numbers.lisp
4208 ** code/early-numbers.lisp -> numbers.lisp
4209 ** early-host-type.lisp -> early-type.lisp
4210 ** late-host-type.lisp -> late-type.lisp
4211 ** host-typep.lisp -> typep.lisp
4212 ** load.lisp -> target-load.lisp
4213 ** host-load.lisp -> load.lisp
4214 ** host-disassem.lisp -> disassem.lisp
4215 ** host-insts.lisp -> insts.lisp
4216 ** byte-comp.lisp -> target-byte-comp.lisp
4217 ** host-byte-comp.lisp -> byte-comp.lisp
4218 ** host-signal.lisp -> signal.lisp
4219 ** host-defstruct.lisp -> defstruct.lisp
4220 ** late-target-type.lisp -> deftypes-for-target.lisp
4221 Furthermore, several other previously target-only files foo.lisp (e.g.
4222 hash-table.lisp and random.lisp) have been split into a target-and-host
4223 foo.lisp file and a target-only target-foo.lisp file, with their key type
4224 definitions in the target-and-host part, so that the cross-compiler will
4225 know more about target types.
4226 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4227 gone away. In their place are various *BACKEND-FOO* variables
4228 corresponding to the slots of the old structure.
4229 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4230 into the target SBCL has been fixed.
4231 * The chill.lisp system for loading cold code into a running SBCL
4233 * Support for the CMU CL "scavenger hook" extension has been removed.
4234 (It was undocumented and unused in the CMU CL sources that SBCL was
4235 derived from, and stale in sbcl-0.6.1.)
4236 * Various errors in the cross-compiler type system were detected
4237 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4238 (enabling various consistency checks). Many of them were fixed,
4239 but some hard problems remain, so the compiler is back to
4240 running without *TYPE-SYSTEM-INITIALIZED* for now.
4241 * As part of the cross-compiler type system cleanup, I implemented
4242 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4243 * I have started adding UNCROSS calls throughout the type system
4244 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4245 on cross-compiler input files will be able to go away, and various
4247 * CONSTANTP now returns true for quoted forms (as explicitly required
4250 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4252 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4253 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4254 (MAKE-LIST -1). In the longer term, it ought to become true
4255 that declarations are assertions even at SAFETY 1. For now, it's not
4256 quite true even at SAFETY 3, but it's at least more nearly true..
4257 (Note that this change seems to increases the size of the system by
4258 O(5%) and to decrease the speed of the compiler by 20% or more.)
4259 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4260 for the problem of printing dozens of lines of distracting information
4261 about low-level system machinery as part of the top stack frame
4262 on entry to the debugger when an undefined function was called.
4263 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4264 is not reset to COMMON-LISP-USER.
4265 * Compilation of stuff related to dyncount.lisp has been made conditional
4266 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4267 smaller. The various dyncount-related symbols have been moved into
4268 a new "SB-DYNCOUNT" package.
4269 * tty-inspect.lisp has been renamed to inspect.lisp.
4270 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4271 feature has gone away. (When we eventually port to other flavors of
4272 libc and/or Unix, we'll try to make the differences between flavors
4273 invisible at the user level.)
4274 * Various other *FEATURES* tags, and/or their associated conditionals,
4275 have been removed if obsolescent, or given better documentation, or
4276 sometimes given more-mnemonic names.
4278 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4280 * tidied up "make.sh" script
4281 * tidied up system directory structure
4282 * better "clean.sh" behavior
4283 * added doc/FOR-CMUCL-DEVELOPERS
4284 * many many small tweaks to output format, e.g. removing possibly-confusing
4285 trailing #\. character in DESCRIBE-INSTANCE
4286 * (EQUALP #\A 'A) no longer signals an error.
4287 * new hashing code, including EQUALP hashing
4288 * tidied up Lisp initialization and toplevel
4289 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4290 * command line argument processing
4291 * added POSIX-GETENV function to deal with Unix-ish environment variables
4292 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4293 terminating SBCL on EOF
4294 * non-verbose GC by default
4295 * There is no more "sbcl" shell script; the sbcl file is now the C
4296 runtime executable (just like CMU CL).
4297 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4299 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4300 symbol and package information as Lisp native symbol and package objects
4301 instead of strings naming symbols and strings naming packages. This way,
4302 whenever packages are renamed (as in warm init), debug information is
4303 transformed along with everything else.
4304 * tweaked the optimization policy declarations which control the building
4305 of SBCL itself. Now, among other things, the system no longer saves
4306 source location debugging information. (This helps two problems at once
4307 by reducing SBCL size and by keeping SBCL from trying to look for its
4308 sources -- which may not exist -- when reporting errors.)
4309 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4310 debugging and testing purposes
4311 * cleaned up printing, making the printer call PRINT-OBJECT for
4312 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4313 methods, giving nearly-ANSI behavior
4314 * converted almost all special variables to use *FOO* naming convention
4315 * deleted PARSE-TIME functionality, since it can be done portably
4316 * moved some files out of cold init into warm init
4317 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4319 * regularized formatting of source files
4320 * added an install.sh script
4321 * fixed ridiculous memory usage of cross-compiler by making
4322 compiler/alloc.lisp not try to do pooling unless it can hook
4323 itself into the GC of the cross-compilation host. Now the system
4324 builds nicely on my old laptop.
4325 * added :SB-ALLOC in target-features.lisp-expr
4326 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4327 was not implemented)
4328 * re-did condition handling and note reporting in the compiler. Notes
4329 are no longer handled by signalling conditions. Style warnings
4330 and warnings are handled more correctly and reported in such a way
4331 that it's easy to find one or the other in your output (so that you
4332 can e.g. figure out which of many problems caused COMPILE-FILE to
4334 * changed the severity of several compiler warnings from full WARNING
4335 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4336 compiler note reporting so that it doesn't use the condition system
4337 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4338 COMPILE-FILE command)
4339 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4340 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4341 no longer supported, since their implementation was deeply intertwingled
4342 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4343 * removed bogus "support" for compiler macros named (SETF FOO), and
4344 removed the compiler macro for SETF INFO (but only after making a fool
4345 of myself on the cmucl-imp mailing list by posting a bogus patch for
4346 DEFINE-COMPILER-MACRO..)
4347 * Compiled files containing forms which have side effects on the Lisp
4348 reader (such as DEFPACKAGE forms) are now handled more correctly.
4349 (Compiler queuing of top level lambdas has been suppressed by setting
4350 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4351 * deleted various currently-unused source files, e.g. gengc.lisp. They
4352 may be added back at some point e.g. when porting to other architectures,
4353 but until they are it's distracting to distribute them and to try to
4355 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4356 there were so many of them they're just distractions, and UNCROSS is
4357 known to be able to handle the current sources
4358 * moved PROFILE functionality into TRACE, so that it will be clear
4359 how the wrapping and unwrapping of functions when you profile them
4360 interacts with the wrapping and unwrapping of functions when you
4361 trace them. (Actually, the functionality isn't there yet, but at least
4362 the interface specification is there. Hopefully, the functionality will
4363 arrive with some future maintenance release.)
4364 * removed host-oops.lisp
4365 * changed signature of QUIT function to allow UNIX-CODE argument
4366 * fixed READ-SEQUENCE bug
4367 * tweaked verbose GC output so that it looks more like the progress
4368 output that ANSI specifies for functions like LOAD
4369 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4370 * added <http://sbcl.sourceforge.com> to the banner information printed by