1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
3 * minor incompatible change: the MSI installer on Windows no longer
4 associates .lisp and .fasl files with the installed SBCL.
5 * minor incompatible change: :UNIX is no longer present in *FEATURES*
6 on Windows. (thanks to Luis Oliviera)
7 * new feature: SB-CLTL2 contrib module now implements
8 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
9 * optimization: scavenging weak pointers is now more efficient,
10 requiring O(1) instead of O(N) per weak pointer to identify
11 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
12 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
13 method bodies using constant slot-names when the first argument is
14 a specializer parameter for the method.
15 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
16 names, when the class of the instance is a direct instance
17 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
18 as slow as the constant slot-name case.)
19 * optimization: member type construction is now O(N) instead
21 * optimization: UNION and NUNION are now O(N+M) for large
22 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
24 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
25 now more readable in environments like Slime which display it.
26 (thanks to Tobias C. Rittweiler)
27 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
28 was able to infer the correct argument type for the object on which
29 the CAS operation was being performed.
30 * bug fix: copy propagation interfered with parallel assignment
31 semantics in local calls. (reported by Paul Khuong)
32 * bug fix: the signed modular fixnum shift compiled to wrong code on
33 x86 and x86-64. (spotted by a slight modification to some of
36 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
37 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
38 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
39 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
41 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
42 and generic function now print the object identity for anonymous
43 instances. (thanks to Attila Lendvai)
44 * enhancement: as an extension to MOP, SBCL now supports SETF of
45 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
46 (thanks to Attila Lendvai)
47 * enhancement: sb-cover has an alternate annotation mode which puts
48 the form annotations at the car of the form, instead of annotating
49 it completely. The benefit of this mode is that it shows better
50 which forms were instrumented by the compiler.
51 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
52 by printer control variables. (reported by Dan Corkill)
53 * bug fix: system leaked memory when delayed output was performed by
54 the OS in smaller chunks then expected. (thanks to David Smith)
55 * bug fix: system leaked memory when file streams were not closed
57 * bug fix: large objects written to slow streams that were modified
58 after the write could end up with the modified state written to
59 the underlying file descriptor.
60 * bug fix: multiple threads operating in parallel on the same stream
61 could cause buffer-overflows.
62 * bug fix: source location information is stored correctly
63 (broken since 1.0.6). This bug would generally show up as the
64 Slime debugger higlighting the wrong form when the sldb-show-source
66 * bug fix: Branch forms are again annotated as branches in the sb-cover
68 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
69 on x86-64. (reported by Gregory Vanuxem)
70 * bug fix: EQUALP could return wrong results for structures with raw
71 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
72 word). (reported by Vjacheslav Fyodorov)
73 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
76 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
77 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
78 atomic compare-and-swap operations on threaded platforms.
79 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
80 allows assining a global minimum value to optimization qualities
81 (overriding proclamations and declarations).
82 * enhancement: closed over variables can be stack-allocated on x86
84 * performance bug fix: GETHASH and (SETF GETHASH) are once again
86 * optimization: slot definition lookup is now O(1). This speeds up
87 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
88 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
89 * optimization: ASSOC and MEMBER can now be open-coded for all
90 combinations of keyword arguments when second argument is constant
91 and SPEED >= SPACE. In other cases a specialized version is
93 * bug fix: using obsoleted structure instances with TYPEP and
94 generic functions now signals a sensible error.
95 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
96 (reported by Kristoffer Kvello)
97 * bug fix: backtrace construction is now more careful when making
98 lisp-objects from pointers on the stack, to avoid creating bogus
99 objects that can be seen by the GC.
100 * bug fix: defaulting of values in contexts expecting more than 7
101 variables now works on x86-64. (reported by Christopher Laux)
102 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
104 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
105 as the property-list of a symbol.
106 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
107 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
108 situations requested, are once again file-compileable. (reported
111 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
112 * MOP improvement: support for user-defined subclasses of
113 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
114 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
115 the expansion of DEFMETHOD, is responsible for generating a form
116 which creates a list of specializers when evaluated. Additional
117 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
118 debugging and introspective support.
119 * minor incompatible change: the (unsupported) spinlock interface
120 has changed: free spinlock now has the value NIL, and a held spinlock
121 has the owning thread as its value.
122 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
123 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
125 * enhancement: name of a socket-stream is now "a socket" instead of
127 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
128 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
129 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
130 for generic functions. (Thanks to Tobias C. Rittweiler)
131 * optimization: bignum printing speed has been improved by 20-40%
132 (depending on the bignum size.)
133 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
135 * bug fix: the cache used by the CLOS to store precomputed effective
136 methods, slot offsets, and constant return values is now thread and
138 * bug fix: generic function dispatch function updating is now thread
139 and interrupt safe (in the sense that the known issues have been
141 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
142 * bug fix: interning EQL-specializers is now thread and interrupt safe.
143 * bug fix: asdf systems with dependencies to the SB-POSIX or
144 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
145 * bug fix: interrupt safety of applicable method computation has been
148 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
149 * new contrib: sb-cover, an experimental code coverage tool, is included
151 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
152 significantly faster.
153 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
154 produce more efficient code on x86-64 if the compiler can determine
155 that (+ INDEX <constant>) does not require a bounds check and FOO
156 has an element type at least 8 bits wide.
157 * enhancement: a new, experimental synchronous timeout facility is
158 provided. Refer to SB-SYS:WITH-DEADLINE for details.
159 * enhancement: when a symbol name conflict error arises, the
160 conflicting symbols are always printed with a package prefix.
161 (thanks to Kevin Reid)
162 * enhancement: stepping is now once again supported on the SPARC and
163 MIPS platforms. (It is also now more likely to work on CheneyGC
165 * enhancement: sb-sprof can now also track and report accurate call
167 * bug fixes: the treatment of non-standard subclasses of
168 SB-MOP:SPECIALIZER is more correct.
169 * incompatible change: PURIFY no longer copies the data from the
170 dynamic space into the static and read-only spaces on platforms
171 that use the generational garbage collector
172 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
174 * bug fix: GC race condition occasionally resulting in crashes with
175 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
177 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
179 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
180 a STYLE-WARNING to be signalled (regression from 1.0.4.)
181 * bug fix: an asynchronous interrupt could previously leave the
182 system running with GC inhibited.
183 * bug fix: a DECLARE form evaluated at top-level now causes an error
184 rather than silently (or verbosely) returning NIL.
185 * bug fix: trying to dissassemble functions compiled at high DEBUG could
186 result in a error being signalled due to source form lookup errors.
187 (reported by Peter Graves)
189 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
190 * incompatible change: removed writer methods for host-ent-name,
191 host-ent-addresses -- changing the values did not update the DNS
193 * minor incompatible change: changed experimental JOIN-THREAD interface
194 * documentation: the manual now lists reader and writer methods
195 in class slot documentation sections. (thanks to Richard M Kreuter)
196 * documentation: unwinding from asyncronous events has been
197 documented as unsafe.
198 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
199 in multithreaded application code.
200 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
201 platforms. (thanks to James Anderson for the optimization hint)
202 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
204 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
205 variants no longer cons.
206 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
207 their NOT- variants no longer cons.
208 * optimization: EQUAL hash tables no longer use SXHASH for objects
209 of all data types, but instead use an EQL hash for types for which
210 EQUAL is the same as EQL.
211 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
212 are significantly faster.
213 * optimization: new STRING-OUTPUT-STREAM implementation is much
214 faster for large outputs and conses 30% less on average.
215 * enhancement: XREF information is now collected for references made
216 to global variables using SYMBOL-VALUE and a constant argument.
217 * enhancement: SIGINT now causes a specific condition
218 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
219 * bug fix: FILE-POSITION used to signal an error in some cases where
220 ANSI requires it to return NIL.
221 * bug fix: ADJUST-ARRAY is now interrupt-safe.
222 * bug fix: adding and removing fd-handlers is now interrupt-safe.
223 * bug fix: inlined calls to C now ensure 16byte stack alignment on
225 * bug fix: bad type declaration in the CLOS implementation has
226 been fixed. (reported by James Anderson)
227 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
228 (thanks to Lutz Euler)
229 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
230 required by the CLHS. (reported by Samium Gromoff)
231 * bug fix: dead unbound variable references now signal an error.
232 (reported by Marco Monteiro)
233 * bug fix: / with an unused value was being deleted in safe code.
234 (thanks to Marco Monteiro and Kevin Reid)
235 * bug fix: number of characters that can be written onto a single
236 line in a file is unlimited.
237 * bug fix: some GC deadlocks caused by asynchronous interrupts have
238 been fixed by inhibiting interrupts for when GC is disbled.
239 * bug fix: some interrupt safety issues with GETHASH, (SETF
240 GETHASH), CLRHASH and REMHASH have been fixed.
241 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
242 specifier no longer causes infinite recursion.
243 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
244 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
245 * bug fix: an &environment argument with the correct variable information
246 is passed to macros that are expanded during byte compilation.
247 (reported by Samium Gromoff)
248 * bug fix: a logic error was causing the hash function for bignums to
249 have excessive amounts of collisions. (reported by Faré Rideau)
250 * bug fix: modifying the contents of an array could change the return
251 value of SXHASH on that array, which is only allowed for strings
252 and bit vectors. (bug introduced in 0.9.16)
253 * improvement: the x86-64/darwin port now passes all tests (except
254 for the debugger tests) but should still be considered
255 experimental until this is fixed.
256 * improvement: a style-warning is signaled for CASE (etc) clauses with
257 duplicate keys. (thanks to Kevin Reid)
258 * improvement: macos/x86 and macos/x86-64 now use mach exception
259 handlers for dealing with illegal instructions (for trapping and
260 error handling) and memory protection violations (for GC).
262 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
263 * new platform: experimental support for x86-64/darwin (MacOS).
264 * incompatible change: the thread-safe (on most platforms) getaddrinfo
265 and getnameinfo sockets functions are used instead of gethostbyaddr
266 and gethostbyname, on platforms where the newer functions are available.
267 As a result, the ALIASES field of HOST-ENT will always be NIL on these
269 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
270 don't print the startup banner, but behave as if --noinform was passed
271 as a command line argument. (thanks to Kevin Reid)
272 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
273 * optimization: code using alien values with undeclared types is much faster.
274 * optimization: the compiler is now able to open code SEARCH in more cases.
275 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
276 * bug fix: using standardized COMMON-LISP special variables as loop
277 variables no longer signals bogus package lock violations. (reported
279 * bug fix: declaring local loop variables to be of a range-limited type
280 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
281 (reported by Andras Simon)
282 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
283 bugs remain on x86-64.)
284 * bug fix: the #= and ## reader macros now interact reasonably with
285 funcallable instances.
286 * bug fix: type-checks for function arguments were compiled using the
287 compiler policy settings of the wrong lexical environment (in 1.0.2
289 * bug fix: SHADOW accepts characters as string designators, as required
290 by the spec (thanks to Eric Marsden)
291 * bug fix: fixed GC safety issues when foreign functions are called with
292 non-base strings as arguments
293 * bug fix: more consistent error output for fatal-compiler-warnings, like
295 * bug fix: functions with proper names were showing up as NIL in
296 backtraces (reported by Edi Weitz, regression in 1.0.2)
298 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
299 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
300 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
301 produce more efficient code on the x86 if the compiler can determine
302 that (+ INDEX <constant>) does not require a bounds check and FOO
303 has an element type at least 8 bits wide.
304 * bug fix: references to undefined variables are handled the same way
305 in toplevel forms as in the normal compiler
306 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
308 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
309 (thanks to Jon Buller)
310 * improvement: faster compilation times for complex functions
311 * improvement: added readlink support to SB-POSIX (thanks to Richard
314 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
315 * improvement: experimental support for mach exception handling on
316 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
318 * improvement: support for GBK external format.
319 (thanks to Chun Tian (binghe))
320 * improvement: the debugger now displays variables that have been closed
321 over, in code compiled with (DEBUG 2) or higher
322 * improvement: support for executable cores on NetBSD (thanks to
324 * new feature: added a RESTART-FRAME debugger command
325 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
326 be used to provide an implementation for FILE-POSITION on Gray streams
327 (thanks to Eric Marsden)
328 * new feature: add syslog, openlog and closelog support to SB-POSIX
329 (thanks to Richard Kreuter)
330 * optimization: the function call overhead in code compiled with
331 a high DEBUG optimization setting is significantly
332 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
333 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
334 (thanks to Tony Martinez)
335 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
336 evaluated once (reported by Kevin Reid)
337 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
338 works again (thanks to Kilian Sprotte)
339 * bug fix: an error is signaled for tagbodies with duplicate tags
340 (thanks to Stephen Wilson)
341 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
342 * bug fix: Win32 port can now handle foreign code unwinding Lisp
343 stack frames from alien callbacks.
344 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
345 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
346 65k pages (thanks to David Woodhouse)
347 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
349 changes in sbcl-1.0.1 relative to sbcl-1.0:
350 * new platform: FreeBSD/x86-64, including support for threading.
351 * new feature: the compiler stores cross-referencing information
352 abount function calls (who-calls), macroexpansion (who-macroexpands)
353 and special variables (who-binds, who-sets, who-references) for code
354 compiled with (< SPACE 3). This information is available through the
355 sb-introspect contrib.
356 * new feature: users may subclass SEQUENCE, and have instances of
357 these classes interoperate with standard Common Lisp functions if
358 a number of methods are defined. (This feature is experimental
359 and the interface subject to change based on feedback from SBCL
360 users and the general community)
361 * improvement: sb-sprof traces call stacks to an arbitrary depth on
362 x86/x86-64, rather than the previous fixed depth of 8
363 * improvement: another pthread back-end of mutex "pthread-futex".
364 * bug fix: non-ascii command-line arguments are processed correctly
365 (thanks to Yaroslav Kavenchuk)
366 * bug fix: non-required arguments were not passed correctly when a method
367 defined using DEFMETHOD was called from a mop-generated method using
368 CALL-NEXT-METHOD (reported by Pascal Costanza)
369 * bug fix: recursion is now permitted in accessors through
370 SLOT-UNBOUND. (reported by Pascal Costanza)
371 * bug fix: an error was signaled at startup if the HOME environment
372 variable was defined, but had an empty value (reported by Peter Van Eynde)
373 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
374 signal a PROGRAM-ERROR, not a BUG.
375 * bug fix: SB-POSIX:READDIR works when built with large file support.
376 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
378 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
379 for better automatic indentation support. (thanks to Matt Pillsbury)
380 * optimization: loading generic functions no longer takes O(n^2) time,
381 proportional to the amount of methods in the generic function
382 (reported by Todd Sabin and Jeremy Brown)
383 * optimization: the FIND and POSITION family of sequence functions
384 are significantly faster on arrays whose element types have been
386 * improvements to the Windows port:
387 ** Intermittent heap corruption problems have been fixed. (thanks
388 to Alastair Bridgewater)
389 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
391 ** Lisp is able to unwind foreign exception frames from alien
392 callbacks. (thanks to Alastair Bridgewater)
394 changes in sbcl-1.0 relative to sbcl-0.9.18:
395 * improvement: experimental support for threading on FreeBSD/x86.
396 (thanks to NIIMI Satoshi)
397 * improvement: runtime option --dynamic-space-size can be used
398 to set the size of the dynamic space reserved on startup.
399 * improvement: floating point modes in effect are now saved in
400 core, and restored on startup.
401 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
402 startup, not time since first call to GET-INTERNAL-REAL-TIME.
403 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
404 threads are not running after *SAVE-HOOKS* have run.
405 * improvement: writes to CLOS instance slots are type-checked in code
406 compiled with (SAFETY 3)
407 * improvement: floating-point exception handling on FreeBSD (thanks to
409 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
410 (thanks to Zach Beane)
411 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
413 * improvement: added support for the Shift-JIS external format.
414 (contributed by NIIMI Satoshi)
415 * improvement: callbacks are supported on Linux/PPC. (thanks to
417 * bug fix: compiler bug triggered by a (non-standard) VALUES
418 declaration in a LET* was fixed. (reported by Kaersten Poeck)
419 * bug fix: file compiler no longer confuses validated and already
420 dumped structurres. (reported by Kaersten Poeck)
421 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
422 fill-pointer signals a type-error as required. (thanks to
424 * bug fix: disassemly of funcallable instances works.
425 * bug fix: single stepping on PPC.
426 * bug fix: fix thread-safety problems in the type system (generally
427 manifesting as nonsensical errors like "STRING is a bad type specifier
428 for sequences" or "The value 1 is not of type FIXNUM")
429 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
430 (reported by Josip Gracin)
431 * bug fix: an error is signaled for attempts to displace arrays with
432 incompatible element types (thanks to Mario Mommer)
433 * bug fix: more correct handling of wide characters in debug info
434 (bug reported by Attila Lendvai and fixed by Juho Snellman)
435 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
436 and don't cause extra consing
437 * optimization: MAP and MAP-INTO are significantly faster on vectors
438 whose elements types have been declared.
439 * Improvements to SB-SPROF:
440 ** Support for allocation profiling
441 ** Reduced profiling overhead, especially for long profiling runs
442 * Improvements to the Windows port:
443 ** floating point exceptions are now reported correctly.
444 ** stack exhaustion detection works partially.
445 ** more accurate GET-INTERNAL-REAL-TIME.
446 ** better RUN-PROGRAM behaviour in case of errors in the child
448 ** PROBE-FILE now simplifies pathnames correctly.
449 ** DIRECTORY now works correctly with :WILD-INFERIORS.
450 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
451 1st 1970 (although time timezone data will be unreliable).
452 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
454 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
455 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
456 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
458 * bug fix: two potential GC deadlocks affecting threaded builds.
459 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
460 returns T (reported by Anton Kazennikov)
461 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
462 with non-variable places
463 * bug fix: remove a race condition in the setting of
464 funcallable-instance functions, this should make threaded CLOS
465 code more stable against memory faults.
466 * bug fix: corruption of specials when unbinding is interrupted by an
467 asynchronous unwind (reported by Hannu Koivisto)
468 * improvement: the debugger will now also display local variables that
469 are only used once, for code compiled with a DEBUG optimization quality
472 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
473 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
474 * incompatible change: External-format support for FFI calls. The
475 SB-ALIEN:C-STRING no longer implies an ASCII
476 external-format. Instead the string is subject to external-format
477 conversion. Additionally return values of type SB-ALIEN:C-STRING
478 are by default of type (SIMPLE-ARRAY CHARACTER), not
479 SIMPLE-BASE-STRING. For an alien type that behaves like the old
480 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
481 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
482 * incompatible change: SB-EXT package no longer contains the
483 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
484 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
486 * incompatible change: the single-stepper is no longer available
487 on Alpha, Mips and Sparc platforms.
488 * minor incompatible change: the direct superclasses of
489 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
490 not (STANDARD-OBJECT FUNCTION). This makes the
491 class-precedence-lists of GENERIC-FUNCTION and
492 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
494 * minor incompatible change: the default stream external format on
495 non-unicode SBCL is no longer detected from the locale, but is always
497 * new feature: Add a version of evaluator that uses an interpreter instead
498 of the compiler. EVAL still uses the compiler by default, to switch it
499 to use the interpreter, set the value of the variable
500 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
501 * minor incompatible change: the single-stepper REPL has been merged
502 with the normal debugger (see the "Stepping" heading of the debugger help
503 for more details). The debugger command STEP will no longer switch
504 to the single-stepper REPL.
505 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
506 for a type now works.
507 * bug fix: loading of default sysinit file works. (thanks to Leonid
509 * bug fix: better detection of circularities in the file-compiler.
510 (reported by Marco Monteiro)
511 * bug fix: the CL pathname functions now work with files that have
512 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
513 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
514 systems with Unix98 pty semantics.
515 * bug fix: ASDF-INSTALL will now work with bsd tar.
516 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
518 * bug fix: timers expiring in dead threads no longer cause a
519 type-error (reported by Paul "Nonny Mouse").
520 * bug fix: thanks to more lightweight single-stepper instrumentation,
521 code compiled with (DEBUG 3) will compile and execute significantly faster,
522 and will have more accurate type-inferencing than before
523 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
524 whose bindings are modified
525 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
526 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
527 ** replace dummy implementations of CL:MACHINE-INSTANCE and
528 CL:SOFTWARE-VERSION with proper version
530 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
531 * feature: implemented the READER-METHOD-CLASS and
532 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
533 as specified by AMOP.
534 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
536 * optimization: faster LOGCOUNT implementation on x86 and x86-64
537 (thanks to Lutz Euler)
538 * optimization: hashing of general arrays and vectors has been
539 improved. (reported by Any Fingerhut)
540 * enhancement: SB-INTROSPECT is now able to find definitions of
541 profiled functions. (thanks to Troels Henriksen)
542 * enhancement: compiler-macro expansion applies now to FUNCALL forms
543 as well, allowing compiler-macros for SETF-functions to expand.
544 * enhancement: step-instrumentation no longer wraps known
545 single-value functions with multiple-value context, allowing
546 better type inference.
547 * fixed bug #337: use of MAKE-METHOD in method combination now works
548 even in the presence of user-defined method classes. (reported by
549 Bruno Haible and Pascal Costanza)
550 * fixed bug #339(c): if there are applicable methods not part of any
551 long-form method-combination group, call INVALID-METHOD-ERROR.
552 (reported by Bruno Haible)
553 * fixed bug #361: the :FUNCTION initarg in the protocol for
554 initialization of methods can now be used to override
555 internally-produced optimized functions. (reported by Bruno
557 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
558 system-provided lambda expression no longer cause warnings about
559 unbound #:|pv-table| symbols.
560 * bug fix: improved the handling of type declarations and the
561 detection of violations for keyword arguments with non-constant
563 * bug fix: potentially erronous calls to PATHNAME and
564 MERGE-PATHNAMES were being flushed in some cases.
565 (reported by Richard Kreuter)
566 * bug fix: compiled calls to TYPEP were mishandling obsolete
567 instances. (reported by James Bielman and Attila Lendvai)
568 * bug fix: format strings with ~> without matching ~< no longer
569 trigger an AVER, but signal an understandable error instead.
570 (reported by Antonio Martinez)
571 * bug fix: specifying an output-file without a pathname-type for
572 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
573 of input-file instead of "fasl". (reported by Robert Dodier)
574 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
575 (reported by James Y Knight).
576 * bug fix: compiler-macros-function did not consider the environment
577 argument for shadowing by local functions.
578 * bug fix: compiler-macros expansion was inhibited by local INLINE
580 * bug fix: inline expansions of known functions were subject to
581 step-instrumentation in high debug policies, leading to problems
583 * bug fix: compiler failed to differentiate between different CONS
585 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
586 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
587 * bug fix: #368: incorrect use of expressed vs. upgraded array
589 * bug fix: #306a: more precise unions of array types.
590 * thread-safety improvements:
591 ** CONDITION-WAIT could return early on Linux, if the thread was
592 interrupted and subsequently continued with SIGCONT.
593 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
595 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
596 * added support for the ucs-2 external format. (contributed by Ivan
598 * minor incompatible change: pretty printing of objects of type
599 (cons symbol) is, in the default pprint-dispatch-table, now
600 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
602 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
603 called later in a class's lifetime, possibly as late as when the
604 first instance of the class is created. Previously,
605 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
606 class became finalizeable.
607 * fixed bug: FILE-POSITION sometimes returned inconsistent results
608 for multibyte external-format streams. (thanks to "vbzoli")
609 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
610 with :ALLOCATION :CLASS inherited from superclasses of the
612 * fixed bug: anonymous classes can now be created using the :NAME
613 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
614 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
615 * fixed bug: core-files saved with :EXECUTABLE T can again be
616 executed when SBCL_HOME isn't set. (reported by James Knight)
617 * fixed bug: toplevel LOCALLY forms with declarations could
618 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
619 * fixed bug: printing from several different threads using different
620 values of *print-case* could cause invalid output, due to
621 some internal special variables of the printer not being bound
622 thread-locally (reported by Max Mikhanosha)
623 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
624 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
625 (reported by Pascal Costanza)
626 * fixed bug: FINALIZE-INHERITANCE is called from
627 REINITIALIZE-INSTANCE on classes when the class has previously
628 been finalized, as required by AMOP.
629 * minor code generation optimizations:
630 ** better register allocation in CLOS dispatching functions
631 ** overflow detection when coercing signed bytes to fixnums on x86-64
632 ** is now implemented with one IMUL instruction instead of three shifts
633 ** more efficient bit-vector access on x86 and x86-64
634 ** more efficient access to raw structure slots on x86 and x86-64
635 * fixed some bugs revealed by Paul Dietz' test suite:
636 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
639 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
640 * feature: thread support on Solaris/x86, and experimental thread support
642 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
644 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
645 default method for SLOT-UNBOUND.
646 * minor incompatible change: prevent the user from specializing the
647 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
648 somewhat counter to the intent of the protocol, I (CSR) think, and
649 additionally it just doesn't work in SBCL as currently
650 implemented, thanks to optimizations (that are always valid for
651 the other three SLOT-VALUEish functions, but not for the setter).
652 * minor incompatible change: the :SB-LDB feature is now enabled by
653 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
654 the low-level debugger.
655 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
656 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
657 down to an SB-SYS:FD-STREAM, allowing direct communication in
658 more cases. Temporary files and pipes are still used as a fallback
660 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
661 on the non-reentrant C functions localtime(3) and gmtime(3).
662 * bug fix: global optimization policy was not visible in LOCALLY and
664 * bug fix: class objects can be used as specializers in methods.
665 (reported by Pascal Costanza)
666 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
667 with a type error. (reported by blitz_ on #lisp)
668 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
669 failed with a type error. (reported by Pascal Bourguignon)
670 * bug fix: merging pathnames against defaults with :DIRECTORY
671 starting with '(:RELATIVE :BACK) should preserve the :BACK.
672 (reported by James Y Knight)
673 * bug fix: saving large (>2GB) cores on x86-64 now works
674 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
675 constant index and a (simple-array (signed-byte 32)) array
676 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
677 error (patch by Robert J. Macomber)
678 * feature: TIME now displays the amount of run-time spent in GC
679 * bug fix: The debugger now does a better job of respecting
680 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
681 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
682 workaround for bug 403.)
683 * bug fix: floating point exception handling now works on FreeBSD
684 (thanks to NIIMI Satoshi)
685 * fixed some bugs revealed by Paul Dietz' test suite:
686 ** MISC.641: LET-conversion were not supposed to work in late
688 * improvements to the Win32/x86 port:
689 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
690 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
692 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
693 * new feature: source path information is generated for macro-expansion
694 errors for use in IDE's like Slime (thanks to Helmut Eller)
695 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
696 * bug fix: compiler does not loop forever on an invalid type in
698 * improvement: compilation of most CLOS applications is significantly
700 * optimization: added a limited bytecode compiler for simple toplevel
701 forms, speeding up compilation and FASL loading
702 * bug fix: the statistical profiler now properly distinguishes anonymous
705 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
706 * minor incompatible change: in sbcl-0.9.11 (but not earlier
707 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
708 into subclasses of the finalized class. Now user calls to
709 FINALIZE-INHERITANCE finalize just the one class, and calls by the
710 system return before any subclasses are finalized.
711 * minor incompatible change: The reader no longer ignores errors
712 regarding non-existent packages in #+ and #- feature tests.
713 * new feature: command line options --no-sysinit, --no-userinit to
714 inhibit loading the corresponding init files
715 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
716 for sending data through UDP sockets (thanks to François-René Rideau)
717 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
718 error is signalled instead (thanks to François-René Rideau)
719 * minor incompatible change: Error signalling behaviour of lexical
720 operations violating package locks has changed slightly. Refer to
721 documentation on package locks for details.
722 * enhancement: EVAL can process IF-expressions without resorting to the
724 * bug fix: Some operations involving SIN, COS, or TAN weren't being
725 constant-folded properly due to missing out-of-line functions.
726 (reported by Mika Pihlajamäki)
727 * bug fix: LISTEN sometimes returned T even in cases where no data was
728 immediately available from the stream
729 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
730 were reversed. (reported by Levente Mészáros)
731 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
732 when given a vector displaced to an adjustable array. (reported
734 * bug fix: derivation of float boundaries from numbers outside the
735 appropriate float range (reported by John Wiseman)
736 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
738 * bug fix: precomputing cache entries for generic functions with
739 some subclasses of specializers as yet invalid does not attempt to
740 fill a cache line with a negative offset. (reported by Levente
742 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
743 allow condition class objects as arguments to DOCUMENTATION and
744 (SETF DOCUMENTATION); only find and set documentation for
745 structure names for the STRUCTURE doc-type. (suggested by Gary
747 * improvements to the Win32/x86 port:
748 ** (user-homedir-pathname) and default initialization file
749 locations now know about the user's "Documents and Settings"
750 directory (thanks to Yaroslav Kavenchuk)
751 ** run-program is implemented (thanks to Mike Thomas)
752 ** sockets support (thanks to Timothy Ritchey)
753 ** better backtrace support (thanks to Alastair Bridgewater)
754 ** sb-grovel supported
755 ** asdf-install and sb-posix work somewhat
756 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
757 * improvements to the Solaris/x86 port:
758 ** works on Solaris 11/Solaris Express
759 ** floating-point exception handling support
760 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
761 * fixed some bugs revealed by Paul Dietz' test suite:
762 ** REMOVE-METHOD returns its generic function argument even when
763 the method is not one of the generic functions' methods.
764 ** objects declared as MEMBER types can be admissible arguments to
766 ** printing characters should simply be printed by the FORMAT ~:C
768 ** compiler failure when compiling functions with hairy constant
769 defaults for optional parameters.
770 ** compiler produces wrong code when MAYBE-INLINE-expanding a
771 function, which is already optimized.
773 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
774 * new platform: experimental support for SBCL x86/Darwin, including
775 MacOS X 10.4.5 on Intel.
776 * new feature: Unicode character names are now known to the system
777 (through CHAR-NAME and NAME-CHAR).
778 * new feature: the filesystem location of SBCL's core file is
779 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
780 * minor incompatible change: the contrib modules SB-POSIX and
781 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
782 this change is to make it easier to distribute
783 location-independent binaries.
784 * bug fix: as implied by AMOP, standardized classes no longer have
785 slots named by external symbols of public packages. (reported by
787 * optimization: calling functions via constant symbols -- as in
788 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
789 via the function object as in (FUNCALL #'FOO).
790 * optimization: on x86 and x86-64, the calling convention for the
791 case of dealing with an unknown number of values has been altered
792 to be friendlier to the prediction heuristics implemented,
793 particularly with reference to CALL and RET pairing. (thanks to
794 Alastair Bridgewater)
795 * optimization: on x86, the code for access to thread-local symbol
796 values and binding/unbinding of thread-local symbols is smaller.
797 (thanks to Alastair Bridgewater)
798 * enhancement: CONSTANTP is now able to determine constantness of
799 more complex forms, including calls to constant-foldable standardized
800 functions and some special forms beyond QUOTE.
801 * fixed bug: occasional GC crashes on Solaris/x86
802 * optimization: x86-64 supports stack allocation of results of simple
803 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
804 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
805 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
806 (thanks to James Knight)
807 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
808 by the standard (thanks to Stephen Compall)
810 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
811 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
812 be used for bundling the runtime and the core file into one
813 executable binary. This feature is not currently supported on all SBCL
814 platforms. (thanks to James Bielman and NIIMI Satoshi)
815 * new feature: a generational or ephemeral garbage collector is now
816 the default on the PowerPC platform (both Linux and Darwin). The
817 old Cheney (stop and copy) collector is a build-time option.
818 (thanks to Cyrus Harmon, after Raymond Toy)
819 * minor incompatible change: the method by which SBCL finds its
820 contributed modules has changed; it no longer relies on symbolic
821 links from an $SBCL_HOME/systems directory, but searches directly
822 in the subdirectories of $SBCL_HOME.
823 * enhancement: the dynamic heap size on the Linux/PPC platform is
824 markedly larger, even using the older Cheney garbage collector.
825 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
826 certain complicated string types. (reported by Gary King)
827 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
828 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
829 * fixed bug: DOCUMENTATION on structures no longer signals an
830 error if no structure type of the right name exists. (reported by
832 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
833 applicable non-standard primary or :AROUND method on
834 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
835 no longer cause unbound variable errors. (reported by Kevin Reid)
836 * fixed bug: metacircle resolution in cases where methods have slots
837 added before the slots from STANDARD-METHOD. (reported by Jean
839 * fixed bug: the Power PC assembly code for calling into lisp
840 sometimes computed its return address wrongly (depending on the
841 15th bit of the address, fixed at link time). This used to
842 manifest itself in a segmentation violation while building PCL.
843 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
844 many others over the years)
845 * fixed bug: ROOM no longer reports silly numbers for stack usage if
846 the stack pointer (treated as a signed integer) is negative.
847 (thanks to Peter van Eynde)
849 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
850 * new platform: experimental support for the Windows operating
851 system has been added. (thanks to Alastair Bridgewater)
852 * fixed several bugs in and robustified the PPC FFI (including
853 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
854 * bug fix: allow non-simple string symbol names (reported by Paul
856 * bug fix: interrupt handling on NetBSD (thanks to Richard M
858 * bug fix: saving a core corrupted callbacks on x86/x86-64
859 * bug fix: closed a loophole in metacircularity detection and
860 grounding in the PCL implementation of CLOS.
861 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
862 (reported by tomppa on #lisp)
863 * bug fix: clear the direction flag on Lisp -> C transitions, as
864 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
865 (reported by Andrej Grozin and Hendrik Maryns)
866 * optimization: major improvements to GC efficiency on GENCGC platforms
867 * optimization: faster implementation of EQUAL
868 * optimization: emit more efficient opcodes for some common
869 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
870 * optimization: several other minor code-generation improvements on x86-64
871 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
873 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
874 * minor incompatible change: (SETF CLASS-NAME) and (SETF
875 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
876 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
877 by ANSI as a generic function, and for consistency (SETF
878 GENERIC-FUNCTION-NAME) is treated likewise.
879 * fixed bug #233.b: make constraint propagation notice when a variable
880 value is changed after it is referenced but before it is used
881 * fixed bug #296: no more arbitrary behaviour from filenames with
882 odd characters as --load arguments. (reported by Adam Warner)
883 * fixed bug #390: :CHARACTER-SET pathname components now work as
884 expected. (reported by Tim Daly Jr)
885 * fixed bug #391: complicated :TYPE intersections in slot
886 definitions no longer cause an error in PCL internals.
887 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
888 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
889 returning the number of octets which would be written to the
890 file-stream. (thanks to Robert J. Macomber)
891 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
892 arguments to the internal stream functions. (thanks to David
894 * bug fix: stack exhaustion is now handled on the x86/SunOS
895 platform. (thanks to Daisuke Homma)
896 * bug fix: more accurate ROOM results on GENCGC platforms
897 * optimization: improved type inference for arithmetic-for
898 index variables in LOOP
899 * optimization: faster floating-point SQRT on x86-64
900 * fixed some bugs revealed by Paul Dietz' test suite:
901 ** DOCUMENTATION returns NIL instead of "" for method combinations
902 that don't have a docstring
904 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
905 * minor incompatible change: (SETF CLASS-NAME) and (SETF
906 GENERIC-FUNCTION-NAME) are no longer generic functions, and
907 therefore are not customizeable by user code (as seems to be at
908 least permitted and maybe required by AMOP). As a consolation,
909 however, the SBCL implementation of these functions now calls
910 REINITIALIZE-INSTANCE as specified by AMOP.
911 * bug fix: slot-definition documentation is propagated as per ANSI
912 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
913 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
914 argument, as do eslotd initialization methods. (from Pascal
915 Costanza's "Closer" project)
916 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
917 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
919 * bug fix: it is now possible to have more than one subclass of
920 STANDARD-GENERIC-FUNCTION without causing stack overflow.
921 (reported by Bruno Haible, Pascal Costanza and others)
922 * bug fix: the dependent update protocol now works for generic
923 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
925 * bug fix: condition-class instances corresponding to
926 DEFINE-CONDITION forms are now created eagerly. (reported by
927 Kalle Olavi Niemitalo on comp.lang.lisp)
928 * bug fix: floating point printing is more accurate in some
929 circumstances. (thanks to Simon Alexander)
930 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
931 merged with *DEFAULT-PATHNAME-DEFAULTS*.
932 * bug fix: callbacks on OS X now preserve stack-alignment.
933 (thanks to Cyrus Harmon)
934 * enhancement: the x86-64 disassembler is much better at
935 disassembling SSE instructions. (thanks to Lutz Euler)
936 * enhancement: CHECK-TYPE on a variable now allows type-inference
937 (thanks to Vincent Arkesteijn)
938 * optimization: improved performance of EUC-JP external format.
939 (thanks to NIIMI Satoshi)
940 * optimization: performance improvements to IO on file streams of
941 :ELEMENT-TYPE CHARACTER
942 * optimization: much faster memory allocation on x86-64
943 * optimization: faster garbage collections (latency and throughput)
945 * optimization: faster bignum subtraction and division on x86 and x86-64
946 (thanks to Lutz Euler)
947 * optimization: more accurate type-inference for loops that use a
948 floating point index variable or a negative step.
950 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
951 * bug fix: add a workaround to SBCL looping infinitely at startup on
952 Linux kernels with apparently buggy implementations of personality().
953 (thanks to Svein Ove Aas)
954 * bug fix: Unicode symbols are correctly printed in LDB backtraces
955 (thanks to David Lichteblau)
956 * bug fix: local bindings shadow global symbol macros for
957 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
958 * optimization: non-open coded uses of numeric comparison operators
959 (e.g. >) no longer cons when called with more than one parameter
960 on platforms supporting dynamic-extent allocation.
961 * enhancement: saving cores with foreign code loaded is now supported
962 on MIPS/Linux in addition to the previously supported platforms.
963 * bug fix: FIND-RESTART now tests for activity, not applicability when given
964 a restart object as identifier. (reported by Helmut Eller for CMUCL)
965 * bug fix: division by zero in sb-sprof when no samples were collected
966 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
967 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
968 (thanks to Alexander Kjeldaas)
969 * new feature: ignoring the return values of destructive functions
970 like NREVERSE signals a compile-time style-warning.
971 (thanks to Kevin Reid)
972 * new feature: the alignment of alien structure fields can be
973 explicitly specified. (thanks to Cyrus Harmon)
974 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
975 and dump core on SIGQUIT
977 ** incompatible change: threads do not inherit values of specials
978 from their parents (see manual)
979 ** bug fix: threads stacks belonging to dead threads are freed by the
980 next exiting thread, no need to gc to collect thread stacks anymore
981 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
982 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
983 ** bug fix: dynamic variable and thread start related gc lossage
984 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
986 * fixed some bugs revealed by Paul Dietz' test suite:
987 ** SUBTYPEP is slightly more accurate on heinously complicated
988 CONS types where some of the members have uncertain (in the
989 NIL, NIL sense) type relationships to each other.
991 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
992 no longer purified unless :PURIFY T is explicitly specified.
993 ** Non-purified cores are significantly smaller than before
995 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
996 * new feature: timers based on Zach Beane's excellent timer package
997 * added support for the following external formats: koi8-u,
998 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
999 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1000 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1001 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1002 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1003 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1004 * incompatible change: a threaded SBCL will no longer revert to
1005 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1006 * bug fix: interrupts are disabled until startup is complete; no
1007 more sigsegvs when receiving a signal to soon
1008 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1010 * bug fix: add a workaround for the memory randomization features in
1011 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1012 workaround will only be in effect on systems with the proc filesystem
1014 * bug fix: printing objects of type HASH-TABLE signals a
1015 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1017 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1018 have been running for over 50 days. (reported by Gilbert Baumann)
1019 * bug fix: the logic for getting names of functions gets less
1020 confused when confronded with alternate-metaclass
1021 funcallable-instances. (reported by Cyrus Harmon)
1022 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1023 consistent, even on internal alternate-metaclass objects.
1024 * bug fix: flush closure information collected by physical
1025 environment analysis prepass before the main pass. (bug reported
1027 * bug fix: compiler pack phase does not modify a hashtable, which is
1028 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1029 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1030 may be expanded. (reported by Alexander <malishev> on c.l.l)
1031 * bug fix: no more style-warnings when compiling code using LAST
1032 under high-SPEED low-SPACE settings. (reported by David Wallin
1034 * fixed some incompatibilities between SBCL's MOP and the MOP
1036 ** the METAOBJECT class is now implemented;
1037 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1038 STANDARD-OBJECT, as required;
1039 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1040 now compatible (as required by VALIDATE-SUPERCLASS); there
1041 remains a consistency requirement over the presence or absence
1042 of the FUNCTION class in the superclass list of finalized
1043 classes; see the manual for more details;
1044 ** the :AROUND method for COMPUTE-SLOTS on
1045 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1046 requested slot ordering.
1048 ** bug fix: parent thread now can be gc'ed even with a live
1050 ** bug fix: binding a special with PROGV to no value is not
1051 the same as that symbol not having been bound (thanks to
1053 ** bug fix: binding specials is thread safe (thanks to
1055 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1056 and SB-SPROF do not die with 'no handler for signal XX in
1057 interrupt_handle_now(..)' anymore
1058 ** bug fix: WITH-TIMEOUT works with multiple running threads
1059 * fixed some bugs revealed by Paul Dietz' test suite:
1060 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1061 the :method-class keyword argument.
1063 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1064 * new port: the Solaris operating system on x86 processors is now
1065 mostly supported, though some rough edges in the environment
1066 remain. (thanks to Daisuke Homma)
1067 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1068 space for its heap. (thanks to Thiemo Seufer)
1069 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1070 closures. (thanks to Thiemo Seufer)
1071 * minor incompatible change: eof selects abort in the debugger.
1072 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1073 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1075 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1076 deprecated, and will go away in a future revision of SBCL.
1077 * minor incompatible change: GC-ON and GC-OFF are no longer
1078 implemented with a counter, it does not matter how many times gc
1079 is switched on or off
1080 * bug fix: discriminating functions for generic function classes
1081 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1082 make invalid assumptions about method precedence order. (reported
1084 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1085 platforms. (thanks to Thiemo Seufer)
1086 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1087 * bug fix: correct pathname printing with printer escaping is on.
1088 (thanks to Kevin Reid)
1089 * bug fix: complex VOP definitions in "user-space" no longer trigger
1090 package locks. (reported by Zach Beane)
1091 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1092 for funcallable-instance functions, and is no different from
1094 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1095 characters in the current readtable. (reported by Nicholas Neuss)
1096 * bug fix: SUBTYPEP on various CONS types returns more a more
1097 accurate acknowledgment of its certainty.
1098 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1099 lists in some cases. This partially fixes bug 384.
1100 * flush all standard streams before prompting in the REPL and the
1102 * bug fix: signal handling and triggering gc do not conflict
1103 directly anymore, in particular a high frequency sb-sprof does
1104 not prevent gc from running
1105 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1106 approximation for timezone and DST information between the
1107 universal time epoch and the smallest negative 32-bit time_t.
1108 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1109 year 1899 so long as the encoded time is non-negative.
1110 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1111 the correct answer for non-constant rotate amounts when performing
1112 an inline 32-bit rotation.
1114 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1115 there is only one thread in the session
1116 ** bug fix: memory leak for streams created in one thread and
1117 written to in another
1118 ** bug fix: lockup when compiled with gcc4
1119 ** bug fix: race that allows the gc to be triggered when gc is
1121 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1122 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1124 ** bug fix: finished threads are gc'ed properly
1125 * fixed some bugs revealed by Paul Dietz' test suite:
1126 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1127 has arguments with a different set of applicable methods from
1128 the orignal arguments.
1129 ** The type error thrown by MAP now has a correct expected-type
1131 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1132 name a compiled function.
1133 ** Three MISC tests where a large bignum was improperly coerced to
1134 a float that couldn't represent that bignum during type
1135 derivation were fixed.
1136 ** SUBTYPEP can now handle types involving the negation of a
1137 list-form FUNCTION type.
1138 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1139 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1140 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1142 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1143 * New feature: Experimental support for bivalent streams: streams
1144 opened with :element-type :default now allow character and binary
1145 (unsigned-byte 8) I/O
1146 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1147 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1149 * bug fix: no more highly sporadic "couldn't check whether ~S is
1150 readable" when reading a stream and an interrupt hits in the middle
1151 of a select system call
1152 * compiler better recognizes complex arrays (reported by Eduardo
1154 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1155 for variables on SBCL built with linkage-tables. (reported by Luis
1157 * various error reporting improvements.
1158 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1159 (Thanks to Thiemo Seufer)
1160 * bug fix: nested reader invokations work correctly
1161 * bug fix: it is possible to have simultaneous references to foreign
1162 code and foreign data with the same name.
1164 ** added x86-64 support
1165 ** incompatible change: the threading api now works with thread
1166 objects instead of thread ids
1167 ** bug fix: threads are protected from signals and interruption when
1168 starting up or going down
1169 ** bug fix: a race where an exiting thread could lose its stack to gc
1170 ** bug fix: don't halt on infinite error in threads if possible
1171 ** fixed numerous gc deadlocks introduced in the pthread merge
1172 ** bug fix: fixed thread safety issues in read and print
1173 ** bug fix: debugger doesn't hang on session lock if interrupted at
1174 an inappropriate moment
1175 ** bug fix: run-program is now thread safe(r)
1176 ** bug fix: inner with-recursive-lock no longer releases the mutex
1177 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1178 * fixed some bugs revealed by Paul Dietz' test suite:
1179 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1180 EXPECTED-TYPE slots filled.
1181 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1183 ** MAKE-CONDITION accepts classes as type-designators.
1184 ** COMPILE may never return NIL.
1185 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1186 range before calling Unix time functions
1188 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1189 * numerous signal handling fixes to increase stability
1190 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1191 * minor incompatible change: we now correctly canonize default
1192 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1193 as per the MOP, rather than the historical (INITARG INITFUNCTION
1194 INITFORM). (reported by Bruno Haible)
1195 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1196 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1197 for more information.
1198 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1199 pathname is a directory pathname.
1200 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1201 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1203 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1204 :METACLASS argument in addition to a class name. (reported by
1205 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1206 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1207 (reported by Bruno Haible; patch from Gerd Moellmann)
1208 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1210 * bug fix: more cleanups to the floating point exception handling on
1211 x86-64 (thanks to James Knight)
1212 * bug fix: the compiler does not try to constant fold calls of
1213 COERCE and BIT-* functions when they return freshly constructed
1214 objects. (reported by Nikodemus Siivola and Paul Dietz)
1215 * optimization: improved the MIPS versions of generic (in the
1216 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1217 * optimization: direct conversion of (unsigned-byte 32) to floats on
1218 the PowerPC platform.
1219 * optimization: structure instances with raw slots now use less
1220 memory, and probably show better memory locality. (thanks to
1222 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1223 are treated as requests for stack allocation on the x86-64,
1224 Alpha-32, PPC and SPARC platforms.
1225 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1226 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1227 (thanks to Tony Martinez)
1229 ** gcing a dead thread can no longer lead to lockups
1230 ** threads block signals until they are set up properly
1231 ** errno is no longer shared by threads
1232 ** interrupt-thread restores the eflags register on x86
1233 ** fixed some lockups due to gc/thread interaction
1234 * fixed some bugs revealed by Paul Dietz' test suite:
1235 ** invalid dotted lists no longer raise a read error when
1236 *READ-SUPPRESS* is T
1237 ** signal an error if a symbol that names a declaration is used
1238 as the name of a type, or vice versa
1239 ** allow using the (declare (typespec var*)) abbreviation for
1240 (declare (type typespec var*)) with all type specifiers
1241 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1242 ** always return NIL from PPRINT-POP when OBJECT is NIL
1243 ** don't signal errors when pretty-printing malformed LABELS,
1244 FLET or MACROLET forms
1245 ** declarations in a DOLIST body are also in scope for the
1247 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1249 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1252 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1253 * fixed cross-compiler leakages that prevented building a 32-bit
1254 target with a 64-bit host compiler.
1255 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1256 opened with :IF-EXISTS :OVERWRITE.
1257 * fixed bug 281: error for an invalid qualifier in a short-form method
1258 combination method is not signalled until the faulty method is called.
1259 * bug fix: iteration variable type inferrer failed to deal with open
1260 intervals. (reported by Alan Shields)
1261 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1262 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1263 or more runtime options were provided to the sbcl binary.
1264 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1265 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1266 * optimization: there's now a fast-path for fixnum arguments in the
1267 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1268 * optimization: the code generated on x86-64 is more compact thanks
1269 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1270 * bug fix: floating-point exception handling now partly works on
1271 x86-64. (Thanks to James Knight)
1272 * improvement to the MIPS backend from Thiemo Seufer: C-style
1273 64-bit long long arguments and return values to alien functions
1275 * fixed some bugs revealed by Paul Dietz' test suite:
1276 ** the type-error signalled from WARN has a filled-in DATUM slot.
1277 ** the type-error required when a stream is not associated with
1278 a file has the stream as its datum.
1279 ** type-errors on single-floats on x86-64 no longer have
1280 :INVALID-OBJECT as the datum
1281 ** the type-errors signalled for invalid function names now have
1282 a correct expected type
1283 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1284 for structures defined with a :TYPE. Documentation strings for
1285 typed structures are no longer immediately discarded
1286 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1287 broadcast streams and synonym streams. FILE-LENGTH now also works
1288 on broadcast streams.
1290 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1291 * incompatible change: the --noprogrammer option, deprecated since
1292 version 0.7.5, has been removed. Please use the equivalent
1293 --disable-debugger option instead.
1294 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1296 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1297 inoperational for a while now) has been completely removed.
1298 * null lexical environments are now printed as #<NULL-LEXENV>,
1299 significantly reducing the amount of clutter in typical backtraces.
1300 * documentation on weak pointers, finalization, and after GC hooks
1301 has been added to the manual.
1302 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1303 as other specialized array types, is much faster. SUBSEQ and
1304 COPY-SEQ on such arrays have also been sped up.
1305 * optimization: EQL is now more efficient when at least other argument
1306 is known to be of type (OR FIXNUM (NOT NUMBER)).
1307 * fixed bug: the runtime is now less vulnerable to changes in the
1308 size of the SBCL object on OS X, and virtual memory is reserved for
1309 all spaces that need to be at a fixed address.
1310 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1311 * fixed bug: finalizers and after GC hooks that cause consing are now
1313 * fixed bug: compiler error messages and summaries are now printed to
1314 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1315 * fixed inference of the upper bound of an iteration variable.
1316 (reported by Rajat Datta).
1317 * fixed bug 211e: calling local functions with duplicated constant
1318 keyword argument no longer causes a bogus style warning about an
1320 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1321 ftype declaration to be disregarded. (reported by Dave Roberts)
1322 * fixed bug 373: caused by erronous compilation of references to alien
1323 variables in the runtime on ppc/darwin.
1324 * fixed bug 376: CONJUGATE type deriver.
1325 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1326 (reported by Baughn on #lisp)
1327 * a cross-compiler bug on non-x86 platforms has been identified and
1328 fixed. (thanks to Bruno Haible)
1329 * improvements to the MIPS runtime code for increased stability.
1330 (thanks to Thiemo Seufer)
1331 * increased the maximimum compact environment size to allow
1332 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1333 * partial workaround for bug 135: don't name the function we're
1334 calling for hairy cases of EVAL, so as not to accumulate one environment
1335 entry per call to EVAL. (reported by Kevin Reid)
1336 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1337 * optimization: type testing for non-vector arrays should be faster.
1338 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1340 * bug fix: setting 31st element of a bit vector to zero did not work
1342 * bug fix: redefining a class definition which failed due to a
1343 previous accessor / function clash now works (but see BUGS entry
1344 #380 for more problems in this area). (thanks to Zach Beane)
1345 * the long form of DEFINE-METHOD-COMBINATION disables method group
1346 checking when given a single method group with pattern *. (thanks
1348 * on x86 compiler supports stack allocation of results of simple
1349 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1350 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1351 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1352 * fixed some bugs related to Unicode integration:
1353 ** the restarts for recovering from input and output encoding
1354 errors only appear when there is in fact such an error to
1356 * fixed some bugs revealed by Paul Dietz' test suite:
1357 ** MISC.549 and similar: late transformation of unsafe type
1358 assertions into derived types caused unexpected code
1360 ** SCALE-FLOAT type deriver is less wrong.
1361 ** type derivers for EXP, LOG and similar functions compute result
1362 types for complex arguments better.
1363 ** (MISC.563) CONJUGATE type deriver works for very restricted
1365 ** out-of-line type testers for character strings are available.
1366 ** EQUAL compiler transform understands specialness of objects
1368 ** accessing double-floats stored on the stack now works on x86-64.
1369 ** debugger internals could sometimes create invalid lispobjs,
1370 resulting in GC crashes.
1371 ** MISC.548: type check weakening can convert required type into
1373 ** initialization forms for bindings are not in scope of free special
1376 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1377 * incompatible change: thread support for non-NPTL systems has
1378 been removed - locking is buggy and unreliable. A threaded
1379 SBCL build will now warn at startup and refuse to create
1380 new threads, unless futex support is detected in the kernel
1381 * incompatible change: the top level REPL now has only an ABORT
1382 restart associated with it, not TOPLEVEL and ABORT as it used to.
1383 TOP and TOPLEVEL are now available as debugger commands for
1384 returning to the top level.
1385 * incompatible change: forms evaluated in the REPL now use the
1386 global optimization policy.
1387 * incompatible change: user- and system-initialization files are
1388 no longer processed with LOAD, but by READ and EVAL; hence the
1389 global optimization policy, startup package, readtable, etc,
1391 * The .fasl file format number has been incremented because of
1392 various incompatible changes.
1393 * internal entry point details and argument counts no longer appear
1394 in backtraces unless explicitly requested by setting
1395 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1396 * built-in and standard functions no longer have names like "top
1397 level local call to FOO".
1398 * fixed bug 32: functions defined in non-null lexical environments
1399 now have more legible printed representation
1400 * fixed bug 33: functions defined in non-null lexical environemnts
1401 are now more amenable to inspection by INSPECT.
1402 * workaround for bug 354: XEPs no longer appear in backtraces unless
1403 explicitly requested.
1404 * fixed bug: receiving the signal which results from attempting to
1405 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1406 SIGBUS on other *BSDs) on architectures where the C stack is also
1407 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1408 to debugger confusion or wild instability. (reported by Cheuksan
1410 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1411 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1412 (reported by Lutz Euler)
1413 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1414 compiled code. (reported by Paul Dietz)
1415 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1416 specialized parameter to a method to something that is not TYPEP
1417 the specializer is now possible.
1418 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1419 face of package deletion.
1420 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1421 pathname. (reported by David Tolpin on c.l.l)
1422 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1423 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1424 STRING-STREAM and FILE-STREAM, respectively.
1425 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1426 than previously; in addition, its test suite is now run on build.
1427 (thanks to Luke Gorrie)
1428 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1429 * optimization: added a immediate representation for single-floats
1431 * fixed some bugs revealed by Paul Dietz' test suite:
1432 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1433 ** attempting to create a package with a colliding nickname causes
1434 correctable errors to be signalled.
1435 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1436 ** several x86-64 backend bugs related to sign-extension of immediate
1439 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1440 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1441 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1443 * fixed loading of multiply forward-referenced layouts.
1444 (thanks to Cheuksan Wang)
1445 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1446 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1447 (see COMPILE-FILE documentation for details of :PRINT option)
1448 * fixed bugs 19 and 317: fixed-format floating point printing is
1449 more accurate. This also fixes a bug reported by Adam Warner
1450 related to the ~@F format directive.
1451 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1453 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1454 dispatch table if the from-char is a dispatch macro character.
1455 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1456 by the wordsize no longer ignore the last word. (reported by Lutz
1458 * fixed bug in type checking of dynamic-extent variables. (reported
1460 * optimization: sequence traversal functions use their freedom to
1461 coerce function designators to functions.
1462 * optimization: code with many calls to CLOS methods specialized on
1463 CLOS classes has had redundant type checks removed.
1464 * optimization: added declarations to speed up operations that access
1465 the internal character database (for example STRING-UPCASE)
1466 * optimization: comparison operations between floats and sufficiently small
1467 fixnums no longer create extra rationals
1468 * fixed some bugs related to Unicode integration:
1469 ** portions of multibyte characters at the end of buffers for
1470 character-based file input are correctly transferred to the
1471 start of the buffer at the next read.
1472 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1473 passing it through to OPEN.
1474 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1475 argument given to internal calls to OPEN.
1476 ** the built-in comment readers (introduced by character sequences
1477 ";" and "#|") are more forgiving to encoding errors; they will
1478 STYLE-WARN and then attempt to resync the stream at a character
1479 boundary. (thanks to Teemu Kalvas)
1480 * fixed some bugs in the x86-64 port:
1481 ** Negative short int return values from c-calls are sign-extended
1483 ** The stack is aligned to 16-bytes for c-calls, as required by
1484 the ABI. (thanks to Cheuksan Wang)
1485 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1486 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1487 for enums. (thanks to Vincent Arkesteijn)
1488 ** Multiple small optimizations and bugfixes for floating point
1490 * fixed some bugs revealed by Paul Dietz' test suite:
1491 ** Space, Tab, Linefeed, Return and Page have the invalid
1492 secondary constituent character trait.
1493 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1495 ** WITH-INPUT-FROM-STRING should only update the index place on
1497 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1498 works more reliably.
1499 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1500 with one-dimensional arrays with fill pointers.
1501 ** TYPE-OF failed on a complex with an integer realpart and a
1503 ** compiler failure during type inference for the code of form
1504 (IF (EQL X (THE ...)) ...) (MISC.535).
1506 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1507 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1508 platform. The port supports 61-bit fixnums, large memory spaces
1509 and reloading shared object files.
1510 * enhancement: saving cores with foreign code loaded is now
1511 supported on ppc/Darwin in addition to the previously supported
1513 * enhancement: the statistical profiler now walks deeper into the
1514 call stack for more meaningful call-graphs and accrued time
1515 reports (x86/x86-64 only). It also now reports time spent in
1517 * enhancement: it is now possible to trace most individual methods
1518 of a generic function in addition to tracing the generic function
1520 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1522 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1523 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1524 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1525 produces an error. (thanks to Vincent Arkesteijn)
1526 * bug fix: NAMESTRING on pathnames with :WILD components in their
1527 directories works correctly. (thanks to Artem V. Andreev)
1528 * fixed bug 125: compiler preserves identity of closures. (reported
1530 * bug fixed: functions with &REST arguments sometimes failed with
1531 "Undefined function" when compiled with (DEBUG 3). (reported by
1533 * bug fix: overflow during compiling of setting element of a bit
1534 vector with constant index and value. (reported by Timmy Douglas)
1535 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1536 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1537 * on x86 compiler partially supports stack allocation of dynamic-extent
1539 * GO and RETURN-FROM do not check the extent of their exit points
1540 when compiled with SAFETY 0.
1541 * fixed some bugs related to Unicode integration:
1542 ** encoding and decoding errors are now much more robustly
1543 handled; it should now be possible to recover even from invalid
1544 input or output to the terminal. (thanks to Teemu Kalvas)
1545 ** provided a first cut at implementing STRING-TO-OCTETS and
1546 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1547 ** altered the SB-MD5 contributed module slightly, changing the
1548 interface just enough for it to be supportable for builds where
1549 lisp characters are not eight bits.
1550 * fixed some bugs revealed by Paul Dietz' test suite:
1551 ** the FORMATTER-generated functions for ~V[ conditionals require
1552 the correct number of arguments.
1553 ** READ-FROM-STRING returns the mandated second value when applied
1554 to displaced strings.
1555 ** the #\Rubout and #\Backspace characters are treated as invalid
1556 constituent characters by the tokenizer.
1558 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1559 * new feature: reloading changed shared object files with
1560 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1561 * new feature: references to foreign variables and functions
1562 can now be compiled and loaded before the corresponding shared
1563 object file is loaded, as long as the foreign definitions are
1564 available at runtime.
1565 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1566 supported. (thanks to Dan Debertin)
1567 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1568 just keywords. (thanks to Vincent Arkesteijn)
1569 * fixed bug #331: structure-class instances corresponding to
1570 DEFSTRUCT forms are now created eagerly.
1571 * fixed bug #345: backtraces from calls to undefined functions work
1572 on x86 as well. Related bug #61 is now also partially fixed on x86
1573 (backtraces from throws to unknown catch tags.)
1574 * bug fix: lambda-list parsing is now stricter vrt. order and number
1575 of lambda-list keywords.
1576 * bug fix: as specified by AMOP, an error is signalled if a
1577 class-option appears multiple times in a DEFCLASS form. (reported
1579 * bug fix: RANDOM can be compiled when the compiler derives the type
1580 of its numeric argument as a disjoint set of small integers.
1581 (reported by Paul Dietz)
1582 * bug fix: starting a core saved with shared objects loaded when
1583 those objects are not available no longer causes threaded SBCL to
1584 hang. (reported by Sean Ross)
1585 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1587 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1588 argument, which must always be NIL. (reported by Kalle Niemitalo)
1589 * bug fix: printing 1.0d+23 no longer results in an error.
1590 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1591 * bug fix: structure slot setters preserve evaluation order. (thanks
1593 * fixed some bugs related to Unicode integration:
1594 ** RUN-PROGRAM can allow its child to take input from a Lisp
1595 stream. (reported by Stefan Scholl)
1596 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1597 locales. (reported by Ken Causey)
1598 * fixed some bugs revealed by Paul Dietz' test suite:
1599 ** INCF, DECF and REMF evaluate their place form as specified in
1601 ** FORMATTER expands ~{ ~} iteration directives with V or #
1602 parameters correctly.
1603 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1604 consequent uses no arguments correctly.
1605 ** the system has a partial understanding of the (COMPLEX RATIO)
1607 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1610 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1611 * new feature: a build-time option (controlled by the :SB-UNICODE
1612 keyword feature, enabled by default) for building the system with
1613 support for the entire 21-bit character space defined by the
1615 * new feature: the system now has rudimentary external-format
1616 support; the primary user-visible change at this time is that
1617 characters with the high bit set (such as lower-case-e-acute) will
1618 print correctly to a terminal in a UTF-8 environment.
1619 * minor incompatible change: BASE-CHAR no longer names a class;
1620 however, CHARACTER continues to do so, as required by ANSI.
1621 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1622 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1623 has been moved to the SB-EXT package (temporarily re-exported from
1625 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1626 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1628 * fixed bug #351: better error-handlind and reporting for malformed
1630 * fixed bug #350: bignum-printing is now more memory-efficient,
1631 allowing printing of very large bignums, eg. (expt 2 10000000).
1632 (reported by Bruno Haible)
1633 * fixed bug #302: better primitive-type selection for intersection
1635 * fixed bug #308: non-graphic characters now all have names, as
1636 required. (reported by Bruno Haible)
1637 * bug fix: redefining a class with different superclasses now correctly
1638 removes it from the direct-subclasses of its previous superclasses.
1639 (reported by David Morse)
1640 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1641 new value now works. (reported by Bruno Haible)
1642 * bug fix: correct canonicalization of multiple non-standard slot
1643 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1644 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1645 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1646 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1647 now exists, an signals an error.
1648 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1649 during macroexpansion for non-iteration variables as well. (reported
1650 by Bruno Haible for CMUCL)
1651 * bug fix: Cyclic structures and unprintable objects in compiler
1652 messages no longer cause errors. (reported by Bruno Haible)
1653 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1654 and READ-FROM-STRING all now return a primary value of NIL if
1655 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1656 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1657 (reported by Bruno Haible for CMUCL)
1658 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1659 do not designate a condition. (reported by Bruno Haible for
1661 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1662 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1663 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1664 permitting resolution in favour of any of the conflicting symbols.
1665 (reported by Bruno Haible for CMUCL)
1666 * bug fix: EQUAL compiler optimizations is less aggressive on
1667 strings which can potentially compare true despite having distinct
1668 specialized array element types.
1669 * bug fix: unit enumerations can be defined without dividing by
1670 zero. (thanks to Vincent Arkesteijn)
1671 * FORMAT compile-time argument count checking has been enhanced.
1672 (report from Bruno Haible for CMUCL)
1673 * a partial workaround for the bug 262: the compiler does not try to
1674 inline-expand a local function doing RETURN-FROM from a deleted
1675 BLOCK. (thanks to Peter Denno for the bug report and to David
1676 Wragg for the simple test case)
1677 * fixed some bugs revealed by Paul Dietz' test suite:
1678 ** RENAME-PACKAGE allows all package designators as new package
1680 ** constraint propagation and lambda variable substitution are
1681 more cautious in dealing with partially deleted code.
1682 ** compiler sometimes lost reoptimization passes.
1683 ** CERROR, given a condition as condition designator, passes its
1684 remaining arguments to the continue format control without
1686 ** Case-altering FORMAT directives work correctly on non-ASCII
1688 ** The REAL type specifier handles bounds outside the floating
1689 point ranges without signalling FLOATING-POINT-OVERFLOW.
1690 ** Functions with IR1-transformations can create intercomponent
1691 references to global functions.
1692 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1694 ** FORMAT ~R treats a nil value for its first parameter correctly.
1696 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1697 * enhancement: saving cores with foreign code loaded is now
1698 supported on x86/NetBSD and sparc/Linux in addition to the previously
1699 supported platforms.
1700 * bug fix: on some platforms repeated installations caused multiple
1701 copies of HTML documentation to be installed -- should not happen
1702 any more. (reported by Stefan Scholl)
1703 * bug fix: parsing self-recursive alien record types multiple times
1704 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1705 original patch by Helmut Eller for CMUCL)
1706 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1707 (thanks to Richard Kreuter)
1708 * bug fix: defining classes whose accessors are methods on existing
1709 generic functions in other (locked) packages no longer signals
1710 bogus package lock violations. (reported by François-René Rideau)
1711 * bug fix: special variables as DEFMETHOD parameters no longer have
1712 associated bogus type declarations. (reported by David Wragg and
1714 * bug fix: read-write consistency on streams of element-type
1715 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1716 * bug fix: redefiniton of the only method of a generic function with
1717 no DEFGENERIC no longer emits a full WARNING. In addition,
1718 redefinition of generic functions with no DEFGENERIC to an
1719 incompatible lambda list now signals an error. (thanks to Zach
1721 * bug fix: DEFGENERIC now works even when there's a function of the
1722 same name in an enclosing lexical environment. (thanks to Zach
1724 * fixed compiler failure, caused by instrumenting code during
1725 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1726 * optimization: added loop analysis and improved register allocation
1727 * fixed some bugs revealed by Paul Dietz' test suite:
1728 ** POSITION on displaced vectors with non-zero displacement
1729 returns the right answer.
1730 ** (SIMPLE-STRING) is a valid type specifier for sequence
1732 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1734 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1735 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1737 ** The justification version of the FORMAT ~< directive treats
1738 non-zero minpad parameter correctly.
1740 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1741 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1742 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1743 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1744 the supported interface.
1745 * new feature: Single-stepping of code compiled with DEBUG 2 or
1746 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1747 * new feature: saving cores with foreign code loaded is now
1748 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1749 Timothy Moore's work for CMUCL)
1750 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1751 parameters to * instead of NIL if no initform is supplied.
1752 (reported by Johan Bockgaard)
1753 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1754 correctly when FUNCALL appears as the car of the form. Note:
1755 despite this FUNCALL forms are not currently subject to
1756 compiler-macro expansion. (port of Raymond Toy's fix for the
1757 same from CMUCL, reported by Johan Bockgaard)
1758 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1759 (thanks for Teemu Kalvas)
1760 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1761 case instead of terminating the iteration (thanks for Julian
1762 Squires, Sean Champ and Raymond Toy)
1763 * bug fix: incorrect expansion of defgeneric that caused a style
1764 warning. (thanks for Zach Beane)
1765 * bug fix: slot accessor effective method computation works properly
1766 for classes with multiple non-standard applicable methods on
1767 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1768 * on x86 compiler supports stack allocation of results of LIST and
1769 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1770 CMUCL implementation by Gerd Moellmann)
1771 * fixed some bugs revealed by Paul Dietz' test suite:
1772 ** FORMAT strings with both the ~<~:;~> form of the justification
1773 directive and pretty-printing directives cause an error.
1775 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1776 * incompatible change: the internal functions
1777 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1778 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1779 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1780 instead of the old functions.
1781 * new feature: on platforms where "dladdr" is available foreign
1782 function names now appear in backtraces. (based on Helmut Eller's
1784 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1785 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1787 * documentation: ASDF manual is now distributed with SBCL.
1788 * bug fix: SBCL can now load its contributed modules with REQUIRE
1789 even if the system-provided entries have been removed from
1790 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1792 * bug fix: backtraces involving undefined functions or assembly
1793 routines are more informative. (thanks to Brian Downing)
1794 * bug fix: mutually referent alien structures now work correctly.
1795 (reported by Rick Taube)
1796 * bug fix: structures defined by WITH-ALIEN can be referred to
1797 within other definitions in the same WITH-ALIEN.
1798 * bug fix: division operators (MOD, TRUNCATE and the like) with
1799 constant zero divisors and integer dividends no longer generate
1801 * bug fix: provide default methods for INPUT-STREAM-P and
1802 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1803 * bug fix: improve the detection and resolution of MOP metacycles
1804 (where effective method computation depends on the generic function
1805 having its effective method computed). (reported by Bruno Haible)
1806 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1807 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1808 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1809 * bug fix: pathnames with relative directory components can be
1810 represented relative to default pathnames.
1811 * optimization: in taking the GCD of bignums, reduce the two bignums
1812 to approximately the same size (using Euclid's algorithm) before
1813 applying the more sophisticated binary GCD. (thanks to Juho
1815 * optimization: COUNT on bitvectors now operates word-at-a-time.
1816 * optimization: ASH with a positive, but not necessarily constant,
1817 (leftwards) shift, when in a modular context, is compiled to a
1819 * fixed some bugs revealed by Paul Dietz' test suite:
1820 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1821 if the corresponding argument is NIL.
1822 ** FORMAT directives accepting a minpad parameter treat negative
1824 ** literal commas as character parameters to FORMAT directives are
1826 ** literal spaces directly after ~<Newline> directives within a
1827 format-logical-block (~:< ~@:>) do not induce :FILL-style
1828 conditional newlines.
1829 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1830 as specified: it no longer includes conditional newlines).
1831 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1833 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1834 :PER-LINE-PREFIX argument does not evaluate to a string.
1836 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1837 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1838 the manual for details; for now, package locks can be disabled by
1839 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1840 if no major problems are found then it is likely that they will be
1841 compiled in unconditionally.
1842 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1843 unsupported operators on all platforms. To load a shared library
1844 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1845 object file, link it into a shared library outside of SBCL and
1846 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1848 * bug workaround: ROOM T and the GC were not working together
1849 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1850 reporting facility aren't preserved. That reporting has been
1851 disabled, so now until and unless someone figures out how to make it
1852 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1853 * minor incompatible change: as threatened around sbcl-0.8.0, the
1854 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1855 The symbols are also exported from SB-PCL for backwards
1856 compatibility, but more so than before SB-PCL should be treated as
1857 an implementation-internal package.
1858 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1860 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1861 It is known as of this release not to work on the Alpha, however.
1862 * fixed bug #167: errors signalled due to illegal syntax in method
1863 bodies are now more legible.
1864 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1865 designators and can hence be used with TYPEP.
1866 * fixed bug #333: CHECK-TYPE now ensures that the type error
1867 signalled, if any, has the right object to be accessed by
1868 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1869 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1870 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1872 * fixed bug #334: programmatic addition of slots using specialized
1873 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1874 and :ALLOCATION :CLASS slots.
1875 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1876 float exponents. (rereported by Peter Seibel)
1877 * fixed a bug: #\Space (and other whitespace characters) are no
1878 longer considered to be macro characters in standard syntax by
1879 GET-MACRO-CHARACTER.
1880 * fixed bug: initialization of condition class metaobjects no longer
1881 causes an instance of the condition to be created. (reported by
1883 * fixed bug: it is now possible to have slots such that
1884 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1885 description is neither :INSTANCE nor :CLASS.
1886 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1887 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1888 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1889 non-standard SLOT-DEFINITION-ALLOCATION values.
1890 * fixed bugs: various race conditions handling exiting threads.
1891 CL-PPCRE's thread torture test now passes.
1892 * fixed arguably-a-bug: GC time proportional to number of threads in
1893 system even when most of them are idle
1894 * optimization: improved performance of BIT and SBIT on bit-vectors.
1895 * .fasl file incompatibility: The fasl file version number has
1896 been incremented because of changes associated with package locks.
1898 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1899 * minor incompatible change: the system no longer provides
1900 optimization hints (conditions which are TYPEP
1901 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1902 mode; these hints are still emitted when higher SPEED optimization
1904 * new contrib module: a sampling profiler (profiling by statistical
1905 sampling, rather than by instrumenting functions) is available as
1906 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1907 * the behaviour of the standard function ED is now customizeable by
1908 third parties through a hook variable: see ED's documentation
1909 string for information on the protocol.
1910 * the compiler no longer emits efficiency notes for (FUNCALL X)
1911 when the type of X is uncertain under default optimization
1913 * fixed bug 276: mutating a binding of a specialized parameter to a
1914 method to something that is not TYPEP the specializer is now
1916 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1917 exhibit the required behaviour on the x86 platform. (thanks to
1918 Peter van Eynde, Eric Marsden and Bruno Haible)
1919 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1920 even for difficult arguments. (reported by Peter Graves)
1921 * fixed bug 141a: the backquote printer now descends quoted
1923 * fixed another bug in backquote printing: no more destructive
1924 modification of the form's list structure. (reported by Brian
1926 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1927 move between its address being taken and the call to
1928 interrupt_thread, fixing a crashing race condition.
1929 * the SB-POSIX contrib implementation has been adjusted so that it
1930 no longer exhibits ridiculously poor performance when constructing
1931 instances corresponding to C structs.
1933 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1934 * minor incompatible change: the sb-grovel contrib now treats C
1935 structures as alien (in the sense of SB-ALIEN) objects rather than
1936 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1937 has implications for memory management of client code
1938 (sb-grovel-returned objects must now be manually managed) and for
1939 type safety (alien objects now have full types).
1940 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1941 used to control emission of compiler diagnostics, rather than the
1942 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1943 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1944 quality should be considered deprecated.
1945 * install.sh now installs the user manual as well
1946 * (not quite a new documentable feature, but worth considering in
1947 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1948 beginnings of a semantically meaningful condition hierarchy is
1949 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1951 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1952 designator as the defaults argument.
1953 * fixed bug: Displaced arrays whose displaced-to array has become
1954 too small now cause ARRAY-DIMENSION to signal an error, providing
1955 for safer bounds-checking. (reported by Bruno Haible)
1956 * fixed bug: DEFCLASS slot definitions with identical :READER and
1957 :WRITER names now signal a reasonable error. (reported by Thomas
1959 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1961 * fixed bug: Invalid slot specification errors now print correctly.
1962 (thanks to Zach Beane)
1963 * fixed bug 320: Shared to local slot value transfers in class
1964 redefinitions now happen corrently from superclasses as
1965 well. (reported by Bruno Haible)
1966 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1968 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1969 improper lists correctly. (reported by Bruno Haible)
1970 * fixed bug 313: source-transform for <fun-name> was erroneously
1971 applied to a call of a value of a variable with name <fun-name>.
1972 (reported by Antonio Menezes Leitao)
1973 * fixed bug 307: The obsolete instance protocol ensures that
1974 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1975 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1976 remove dead unknown-values globs from the middle of the stack.
1977 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1978 easier to resume long computations after using *BREAK-ON-SIGNALS*
1979 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1980 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1981 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1982 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1983 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1984 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1986 * fixed a (fairly theoretical) bug in string printing: if
1987 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1988 does not have array-element-type equal to the most general string
1990 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1991 function name. (thanks to Nikodemus Siivola)
1992 * fixed bug: compile-time format string checker failed on
1993 non-closed ~{. (reported by Thomas F Burdick)
1994 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1995 don't warn on legal loop code involving more than one aggregate
1996 boolean. (thanks to Nikodemus Siivola)
1997 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1998 in keyword position to MAKE-PACKAGE was not being reported as
2000 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2001 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2003 * fixed bugs 280 and 312: the checking for multiple definitions in a
2004 file is less likely to become confused by uses of inline
2006 * fixed bug: the #S reader macro performs the keyword coercion
2007 specified for slot names. (reported by Kalle Niemitalo)
2008 * fixed bug: lambda lists may contain symbols whose names start with
2009 & but are not lambda-list-keywords; their occurrence triggers a
2011 * fixed bug 321: define-method-combination argument lambda lists do
2012 not cause invalid code to be generated when &optional or &aux
2013 variables have default values. (reported by Bruno Haible)
2014 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2015 have CLOS classes; this appears also to have fixed reported
2016 instability in stack exhaustion detection.
2017 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2018 MUFFLE-WARNING when no associated restart is present is now
2020 * optimization: rearranged the expansion of various defining macros
2021 so that each expands into only one top-level form in a
2022 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2024 * optimization: used a previously little-used slot in symbols to
2025 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2027 * fixed some bugs revealed by Paul Dietz' test suite:
2028 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2030 ** Fixed an optimization bug related to inheritance of initforms
2031 from local to shared slots.
2032 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2033 ** CAST optimizer forgot to flush argument derived type.
2034 ** print/read consistency on floats is now orders of magnitude
2035 more likely. (thanks also to Bruno Haible for a similar report
2037 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2038 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2039 for objects of type REAL. Make it so.
2040 ** SXHASH is specified (infelicitously) to respect similarity,
2041 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2042 it so. (thanks to Markus Ziegler)
2043 ** on the Alpha, the compiler succeeds in compiling functions
2044 returning a known number of arguments greater than 63.
2045 ** fixed handling of invalid NIL arguments in keyword position
2047 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2049 ** provided a readably-printable representation for RANDOM-STATE
2051 ** ensured that pathnames, where they have a namestring, always
2052 print using #P"..." syntax.
2054 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2055 * Support for the forthcoming 2.0 version of the NetBSD kernel
2056 running on x86 hardware has been added. (thanks to Perry
2057 E. Metzger most immediately, and others for their past work)
2058 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2059 current release version; SBCL's OpenBSD support had been broken
2060 since about the time of OpenBSD's switch to ELF binary format.)
2061 * [placeholder for DX summary]
2062 ** user code with &REST lists declared dynamic-extent, under high
2063 speed or space and low safety and debug optimization policy.
2064 * The manual has been converted to Texinfo format and the debugger
2065 chapter from the cmucl manual has been added.
2066 * A facility has been added to extract documentation strings from
2067 sbcl and store them as Texinfo-formatted snippets for inclusion in
2068 the manual (via Texinfo's @include directive)
2069 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2070 the test case to Dave Roberts)
2071 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2072 pointers. (reported by Sean Ross)
2073 * bug fix: PROFILE output is printed nicely even for large numerical
2074 values. (thanks to Zach Beane)
2075 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2076 greater than 32 handle EOF correctly.
2077 * bug fix: on X86 an immediate argument of the IMUL instruction is
2078 correctly printed in disassembly. (thanks to Lutz Euler)
2079 * bug fix: class slots in redefined classes preserve their old
2080 values. (thanks to Bruno Haible and Nikodemus Siivola)
2081 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2082 succeeds. (reported by Marco Baringer)
2083 * fixed some bugs revealed by Paul Dietz' test suite:
2084 ** READ-SEQUENCE now works on ECHO-STREAMs.
2085 ** RATIONALIZE works more according to its specification. (thanks
2088 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2089 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2090 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2091 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2092 should matter to you only if you rebind the printer control
2093 variables and then find you want different bindings in the
2094 debugger than in the ordinary execution of your program.)
2095 * The runtime build system has been tweaked to support building
2096 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2097 assembler and linker. (thanks to Nikodemus Siivola)
2098 * Unbound, undefined, undeclared variables now trigger full
2099 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2100 more often programmer error than deliberate exploitation of undefined
2102 * optimization: the hash algorithm for strings has changed to one
2103 that is less vulnerable to spurious collisions. (thanks to Juho
2105 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2106 needless bounds checking. (thanks to Juho Snellman)
2107 * optimization: implemented multiplication as a modular
2108 (UNSIGNED-BYTE 32) operation on the PPC backend.
2109 * fixed some bugs revealed by Paul Dietz' test suite:
2110 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2112 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2113 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2114 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2116 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2118 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2120 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2121 their output stream on EOF from read.
2122 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2123 have been read to end-of-file.
2124 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2126 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2127 description of determination of which consecutive characters
2129 ** Printing the "Space" character with escaping on now yields "#\\ ",
2130 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2131 ** Reading floating-point numbers with *READ-BASE* set to a number
2132 less than 10 works correctly.
2133 ** Reading floating-point numbers with *READ-BASE* set to a number
2134 more than 10 works correctly.
2135 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2136 the readtable currently in effect.
2138 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2139 * minor incompatible change: parsing of namestrings on a physical
2140 (Unix) host has changed; numbers after the final #\. in a
2141 namestring are no longer interpreted as a version field. This is
2142 intented to be largely invisible to the user, except that the
2143 meaning of the namestring "*.*.*" has changed: it now refers to a
2144 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2145 should usually be replaced by
2146 (make-pathname :name :wild :type :wild :version :wild)
2147 with the added benefit that this is more likely to be portable.
2148 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2149 option to OPEN now signals an error if the file being opened
2150 exists; this may have an impact on existing code.
2151 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2152 signals received, on the PowerPC platforms (both Linux and
2153 Darwin). (thanks to Pierre Mai for pointing out the location of
2155 * several fixes on OS X: The system now builds and runs cleanly on
2156 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2157 (thanks to Brian Mastenbrook)
2158 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2159 second-resolution: integer multiples of 1/3600 between -24 and 24.
2160 (thanks to Vincent Arkesteijn)
2161 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2162 type when called with 1 argument; PEEK-CHAR checked type of
2163 PEEK-TYPE only after having read first character from a
2164 stream. (reported by Peter Graves)
2165 * bug fix: the garbage collector now has much better locality
2166 behaviour, and in particular no longer treats all memory as being
2167 exhausted when many small objects point to each other in a deeply
2169 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2170 recognized as being TYPEP their class.
2171 * bug fix: the PUSHNEW documentation string has been corrected.
2172 (thanks to Vincent Arkesteijn)
2173 * bug fix: defaulting of the value for the last of an atypically
2174 large number of multiple values being bound was not being
2175 performed correctly on the Alpha or PPC platforms
2176 * optimization: implemented multiplication as a modular
2177 (UNSIGNED-BYTE 32) operation on the x86 backend.
2178 * optimization: SEARCH on simple-base-strings can now be open-coded.
2179 (see also contrib/compiler-extras.lisp for inspiration for
2180 teaching the compiler about the Boyer-Moore algorithm).
2181 * value, returned by MAX (and MIN) called with several EQUALP, but
2182 not EQL, arguments now does not depend on compiler settings.
2183 * fixed some bugs revealed by Paul Dietz' test suite:
2184 ** in stack analysis liveness information is propagated from
2185 non-local entry points.
2186 ** pathwise CAST removing failed when the CAST node did not start
2188 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2189 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2191 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2192 host is already defined.
2193 ** RENAME-FILE works on streams instead of signalling an internal
2195 ** PEEK-CHAR uses the current readtable when determining whether
2196 or not a character is whitespace.
2197 ** MERGE-PATHNAMES handles the case when the pathname does not
2198 specify a name while the default-pathname specifies a version.
2199 ** Pathnames now stand a better chance of respecting print/read
2201 ** Attempting to use standardized file system operators with a
2202 pathname with invalid :DIRECTORY components signals a
2204 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2206 ** DIRECTORY on logical pathnames is more correct.
2207 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2208 signal a TYPE-ERROR if their argument is not a stream.
2209 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2210 designator argument does not designate a stream.
2211 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2212 examining the synonym.
2213 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2215 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2216 element-type larger than ([UN]SIGNED-BYTE 32).
2218 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2219 * When built with the :SB-FUTEX feature, threaded builds now take
2220 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2221 for faster/more reliable mutex and condition variable support.
2222 * Incompatible change (but one you probably shouldn't have been using
2223 anyway): the interface and code for arbitrating between multiple
2224 threads in the same user session has been redesigned.
2225 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2226 error when called without an explicit environment argument.
2227 (thanks to Nikodemus Siivola)
2228 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2229 confused about their position. (thanks to Adam Warner and Gerd
2231 * bug fix: Pretty printing backquoted forms with unquotations in the
2232 argument list position of various code constructs such as LAMBDA
2233 now works correctly. (reported by Paul Dietz)
2234 * bug fix: Pretty printing unquotations no longer loses all
2235 stream position information.
2236 * optimization: performance of string output streams is now less
2237 poor for multiple small sequence writes.
2238 * optimization: performance of CSUBTYPEP in the presence of complex
2239 expressions involving CONS and NOT many times has been improved.
2240 (reported by Paul Dietz)
2241 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2243 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2244 streams for the REPL, for the debugger, and so on. (thanks to
2246 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2247 * fixed some bugs revealed by Paul Dietz' test suite:
2248 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2249 is ignored if it is too small, rather than propagating through
2250 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2251 ** extremely complex negations of CONS types were not being
2252 sufficiently canonicalized, leading to inconsistencies in
2254 ** VALUES tranformer lost derived type.
2256 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2257 * fixed a bootstrapping bug: the build process no longer assumes
2258 that the various BOOLE-related constants have the same value in
2259 host and target lisps. (noted by Paul Dietz' test suite on an
2260 SBCL binary built from CLISP)
2261 * The system can now be dynamically linked on the MIPS platform,
2262 which enables dynamic loading of foreign code from Lisp. (thanks
2263 to Ralf Baechle for discussions on the MIPS ABI)
2264 * The system now records debugging information for its own source
2265 files in a filesystem-position-independent manner, relative to
2266 the "SYS" logical host.
2267 * fixed a compiler bug: MV-LET convertion did not check references
2268 to the "max args" entry point. (reported by Brian Downing)
2269 * tweaked disassembly notes to be less confident about proclaiming
2270 some instruction as an LRA. (thanks to Brian Downing)
2271 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2272 now each have their own history, command character, and other
2273 characteristics. (thanks to David Lichteblau)
2274 * fixed some bugs revealed by Paul Dietz' test suite:
2275 ** compiler failure in compiling LOGAND expressions including a
2277 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2278 shift greater than 32.
2279 ** FLUSH-DEST did not mark blocks for type check regeneration.
2280 ** HANDLER-CASE failed to accept declarations in handler clauses
2281 in some circumstances.
2283 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2284 * New code in contrib/sb-introspect (still probably not entirely
2285 stable yet) provides some support for smart Lisp development
2286 environments like SLIME.
2287 * The conditions signalled for errors occurring when loading .fasl
2288 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2289 in a way which should help ASDF recover gracefully.
2290 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2291 stuff has been cleaned up. If you code contrib/ stuff, this might
2292 affect you, and you can look at contrib/README, contrib/STANDARDS,
2293 and/or the 0.8.4.27 diff to check.
2294 * In full calls the compiler now does not generate checks for declared
2295 argument types for all arguments.
2296 * various threading fixes
2297 ** and some experimental patches which didn't make it into
2298 the main tree for this release, but which are shipped in
2299 contrib/experimental-thread.patch as a possible fix for some
2300 failures (deadlock, spinning...) in GC-intensive multithreaded
2302 * fixed PPC build problem (source code incompatibility of different
2303 library versions): added offsetof() hackery which attempts to divine
2304 where glibc maintainers put uc_mcontext today
2305 * fixed bug 282: compiler does not trust type assertions while passing
2306 arguments to a full call.
2307 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2308 &OPTIONAL VALUES type parameter.
2309 * fix bug 214: algorithm for noting rejected templates is now more
2310 similar to that of template seletion. (also reported by rydis on
2312 * fixed bug 141b: printing backquoted information readably and prettily
2313 inserts a space where necessary.
2314 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2315 (CHARACTER 10) are now reported as errors, rather than propagated
2316 as unknown types. (reported by piso on #lisp)
2317 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2318 with values NIL and :ERROR. (thanks to Milan Zamazal)
2319 * fixed bug 191c: CLOS now does proper keyword argument checking as
2320 described in CLHS 7.6.5 and 7.6.5.1.
2321 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2322 longer raise an error; further, using a list as a for-as-arithmetic
2323 counter now raises a meaningful error.
2324 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2325 now understood by sequence creation functions such as MAKE-SEQUENCE
2327 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2328 a STRING-INPUT-STREAM.
2329 * compiler enhancement: SIGNUM is now better able to derive the type
2331 * type declarations inside WITH-SLOTS are checked. (reported by
2333 * fixed some bugs revealed by Paul Dietz' test suite:
2334 ** incorrect optimization of TRUNCATE for a positive first
2335 argument and negative second.
2336 ** compiler failure in let-convertion during flushing dead code.
2337 ** compiler failure while deriving type of TRUNCATE on an
2338 interval, containing 0.
2339 ** ASH of a negative bignum by a negative bignum count now returns
2341 ** intersection of CONS types now canonicalizes properly, fixing
2342 inconsistencies in SUBTYPEP.
2344 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2345 * incompatible change: The --disable-debugger command line
2346 option now clobbers the debugger at a more fundamental
2347 level, by redefining #'INVOKE-DEBUGGER instead of by
2348 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2349 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2350 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2351 enter the debugger REPL and then suffer recursive errors
2352 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2353 while under the new system, BREAK in --disable-debugger mode
2354 terminates the system just as an unhandled error would.
2355 * fixed compiler performance when processing loops with a step >1;
2356 * bug fix: DOCUMENTATION now retrieves generic function
2357 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2358 support has been systematized, and now supports the methods
2359 specified by ANSI, along with a default method and a method for
2360 slot documentation. (reported by Nathan Froyd)
2361 * bug fix: effective methods associated with a generic function are
2362 no longer cached over a change of that generic function's method
2363 combination. (reported by Andreas Fuchs)
2364 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2365 names a non-existent directory. (thanks to Andreas Fuchs)
2366 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2367 the PPC platform, lead to stack corruption; this has been fixed.
2368 (reported by Rainer Joswig)
2369 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2370 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2371 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2372 arithmetic in general, is now much more reliable on the PPC
2374 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2375 platform now returns the right answer.
2376 * optimization: restored some effective method precomputation in
2377 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2378 precomputation is now tunable.
2379 * optimization: compiler-internal data structure use has been
2380 reviewed, and changes have been made that should improve the
2381 performance of the compiler by about 20%.
2382 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2383 simple-base-strings and simple-bit-vectors is improved.
2384 * optimization: the optimization of 32-bit logical and arithmetic
2385 functions introduced in version 0.8.3 on the x86 has been
2386 implemented on the mips, ppc and sparc platforms; an
2387 implementation of the same facility, but for 64-bit arithmetic,
2388 has been added for the alpha.
2389 * microoptimization: the compiler is better able to make use of the
2390 x86 LEA instruction for multiplication by constants.
2391 * bug fix: in some situations compiler did not report usage of
2392 generic arithmetic in (SPEED 3) policy.
2393 * bug 145b fix: compiler used wrong type specifier while converting
2394 MEMBER-types to numeric.
2395 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2397 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2398 index argument now works properly on the Alpha platform.
2399 * bug fix: floating point exception treatment on the Alpha platform
2401 * bug fix: FILE-POSITION works much better on string input and
2402 output streams. (thanks to Nikodemus Siivola)
2403 * bug fix: many threading/garbage collection symptoms sorted.
2404 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2405 might be pseudo-atomic.
2406 * internal change: Stopping for GC is now done with signals not
2407 ptrace. GC is now done in whichever thread wanted it, instead of
2409 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2411 * bug fix: result form in DO is not contained in the implicit
2413 * incompatible change: ICR structure is changed; the value part of
2414 CONTINUATION is now called LVAR; corresponding functions are
2415 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2416 * added type deriver for ISQRT (thanks to Robert E. Brown).
2417 * bug fix: better support for loading from the command line when an
2418 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2420 * fixed some bugs revealed by Paul Dietz' test suite:
2421 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2422 ** ROUND and FROUND now give the right answer when given very
2423 small float arguments.
2424 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2426 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2427 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2428 ** LCM should return a non-negative integer.
2429 ** PARSE-INTEGER returned the index of a terminator instead of the
2430 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2431 ** PARSE-INTEGER returned an incorrect index being applied to a
2433 ** LCM with two arguments of 0 returns 0 rather than signalling
2435 ** unsigned addition of a 32-bit constant with the high bit set no
2436 longer causes an internal compiler error on the x86.
2437 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2438 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2439 with negative last argument.
2440 ** byte specifiers with zero size and position no longer cause
2441 an error during type derivation.
2442 ** bignum multiplication on the Alpha platform now returns the
2444 * porting: The system now builds on SuSE AMD64, although it still
2445 generates a 32-bit binary.
2446 * .fasl file incompatibility: The fasl file version number has
2447 been incremented (because of the changes to internal compiler
2448 data structures referred to above).
2450 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2451 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2452 more accurately, on the Darwin kernel running on PowerPC hardware.
2453 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2454 * Compiler code deletion notes now signal a condition of type
2455 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2456 an associated MUFFLE-WARNING restart.
2457 * The compiler now performs limited argument count validation of
2458 constant format strings in FORMAT, and where appropriate in ERROR,
2459 CERROR and WARN. (thanks to Gerd Moellmann)
2460 * New ASDF-INSTALL contrib can be used for automatic download and
2461 installation of third-party Lisp code from CCLAN or other sites
2463 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2464 which forces another thread to execute a function supplied by the
2466 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2467 now accept and act upon their :ELEMENT-TYPE keyword argument.
2468 (reported by Martin Atzmueller, Edi Weitz)
2469 * bug fix: FILE-POSITION now accepts position designators up to
2470 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2471 is the greater. (thanks to Patrik Nordebo)
2472 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2473 Kalle Olavi Niemitalo)
2474 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2475 printable. (reported by Eric Marsden)
2476 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2477 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2478 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2479 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2480 (thanks to Stig Sandoe)
2481 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2482 UNREAD-CHAR now correctly includes the unread character in the
2483 target sequence. (thanks to Gerd Moellmann)
2484 * bug fix in threaded builds: the system can now be suspended and
2485 resumed by shell job control with minimal disruption.
2486 * bug fixes in times and timezones >2038AD
2487 * better handling of "where is GNU make?" problem in build scripts
2488 (thanks to Nikodemus Siivola)
2489 * new optimization: inside a named function any reference to a
2490 function with the same name is considered to be a self-reference;
2491 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2492 optimization quality.
2493 * new optimization on x86: logical functions and + now have
2494 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2495 used when the result is truncated to 32 bits.
2496 * VALUES declaration is partially enabled.
2497 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2498 * fixed some bugs revealed by Paul Dietz' test suite:
2499 ** The system now obeys the constraint imposed by
2500 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2501 types form a lattice under type intersection.
2502 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2503 ** ASSOC now ignores NIL elements in an alist.
2504 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2505 and (1+ MOST-POSITIVE-FIXNUM) answers.
2506 ** The addition of a method with invalid qualifiers to a generic
2507 function does not cause an error to be signalled immediately;
2508 a warning is signalled, and the error is generated only on
2509 calling the generic function.
2510 * changed .fasl file version number, in part to add type codes for
2511 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2512 obscure ANSI requirements
2514 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2515 * fixed bug 148: failure to inline-expand a local function left
2516 garbage, confusing the compiler.
2517 * fixed bugs 3cd: structure slot readers perform type check if the
2518 slot can have an invalid value (i.e. it is either not initialized
2519 or can be written with a less specific slot writer).
2520 * bug fix: the compiler now traps array references to elements off
2521 the end of an array; previously, the bounds checking in some
2522 circumstances could go off-by-one.
2523 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2525 * type declarations for array element types now obey the description
2526 on the CLHS page "Declaration TYPE", as per discussions on
2527 sbcl-help around 2003-05-08. This means that a declaration
2528 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2529 declaration, all references to BAR will be asserted or assumed
2530 (with THE, so dependent on compiler policy) to involve objects of
2531 type FOO. Note that no such declaration is implied in
2532 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2533 * declared types of functions from the "Conditions"
2534 chapter. (reported by Paul Dietz)
2535 * bug fix: CERROR accepts a function as its first argument.
2536 * bug fix: NTH an NTHCDR accept a bignum as index
2537 arguments. (reported by Adam Warner)
2538 * optimization: character compare routines now optimize comparing
2539 against a constant character. (reported by Gilbert Baumann)
2540 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2541 argument now works properly on the MIPS platform.
2542 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2543 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2544 declared to have a functional type, does not cause an error
2546 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2547 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2548 * fixed bug: if last continuation of a deleted block has a
2549 destination, this destination should be deleted too. (reported by
2551 * fixed a bug in the bootstrap process: the host compiler's values
2552 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2553 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2554 test case from Patrik Nordebo)
2555 * improved the ability of the disassembler on the PPC platform to
2556 provide helpful disassembly notes.
2557 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2558 the class in more cases than previously.
2559 * bug fix: FILE-POSITION now understands :START and :END for
2560 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2561 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2562 8) data. (thanks to David Lichteblau)
2563 * bug fix: it is possible to add a method to a generic function
2564 without lambda list.
2565 * bug fix: reader failed to signal END-OF-FILE inside an
2566 object representation. (reported by Nikodemus Siivola)
2567 * fixed some bugs revealed by Paul Dietz' test suite:
2568 ** LAST and [N]BUTLAST should accept a bignum.
2569 ** condition slot accessors are methods.
2570 ** (VECTOR NIL) is a subtype of STRING.
2572 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2573 * minor incompatible change: some nonsensical specialized lambda
2574 lists (used in DEFMETHOD) which were previously ignored now signal
2576 * minor incompatible change: the system is now aware of the types of
2577 variables in the COMMON-LISP package, and will signal errors for
2578 most violations of these type constraints (where previously they
2579 were silently accepted).
2580 * minor incompatible change: COMPILE-FILE now uses the freedom
2581 afforded (ANSI 3.2.2.3) to use derived function types for
2582 functions defined in the same file. This also permits the system
2583 to warn on static type mismatches and function
2584 redefinition. (Currently it does not work with high DEBUG level.)
2585 * minor incompatible change: VALUES declaration is disabled.
2586 * When issuing notes, the compiler now signals a condition of type
2587 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2588 restart for use in user handlers. It is expected that the
2589 COMPILER-NOTE condition will eventually become a condition
2590 supertype to a hierarchy of note types, which will then be
2591 handleable in a similar fashion. However, at the moment, no such
2592 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2593 but it's an implementation detail, not a classification for the
2595 * Changes in type checking closed the following bugs:
2596 ** type checking of unused values (192b, 194d, 203);
2597 ** template selection based on unsafe type assertions (192c, 236);
2598 ** type checking in branches (194bc).
2599 * A short form of VALUES type specifier has ANSI meaning (it has
2600 increased the number of situations when SBCL cannot perform type
2602 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2603 keywords or constants is permissible.
2604 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2605 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2606 classes. (thanks to Antonio Martinez)
2607 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2608 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2609 argument to a pretty-print stream. Instead, it leaves any such
2610 operations to DESCRIBE-OBJECT methods.
2611 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2613 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2614 lambda lists. (thanks to David Lichteblau)
2615 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2616 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2617 (thanks to Gerd Moellmann)
2618 * fixed bug 246: increased compilation speed of long
2619 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2621 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2623 * DEFCONSTANT now throws a condition of type
2624 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2625 constant to a non-EQL value; CONTINUE and ABORT restarts
2626 respectively change and preserve the value.
2627 * fixed bug 63: The code walker, part of the implementation of CLOS,
2628 is now better at handling symbol macros.
2629 * bug fix: There is no longer an internal implementation type named
2630 CL:LENGTH. (reported by Raymond Toy)
2631 * bug fix: In macro-like defining macros/special operators the
2632 implicit block does not enclose the lambda list.
2633 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2634 enclosed in parentheses) are not suitable as type specifiers, and
2635 their use properly signals an error now.
2636 * bug fix: An argument count mismatch for a type specifier in code
2637 being compiled no longer causes an unhandled error at compile
2638 time, but signals a compile-time warning.
2639 * fixed simple vector readable printing
2640 * bug fix: DESCRIBE takes more care over whether the class
2641 precedence list slot of a class is bound before accessing it.
2642 (reported by Markus Krummenacker)
2643 * bug fix: FORMATTER can successfully compile pretty-printer format
2644 strings which use variants of the ~* directive inside.
2645 * bug fix: SEARCH now applies its TEST predicate to the elements of
2646 the arguments in the correct order. (thanks to Wolfhard Buss)
2647 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2648 in type inference. (thanks to Robert E. Brown)
2649 * bug fix: Reading in symbols with an explicit package name of ""
2650 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2651 * fixed some bugs revealed by Paul Dietz' test suite:
2652 ** NIL is now allowed as a structure slot name.
2653 ** Arbitrary numbers, not just REALs, are allowed in certain
2654 circumstances in LOOP for-as-arithmetic clauses.
2655 ** Multiple class redefinitions before slot access no longer
2656 causes a type error.
2657 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2658 association between the name and a class.
2659 ** Generic functions with non-standard method-combination and over
2660 five methods all of which return constants no longer return NIL
2661 after the first few invocations. (thanks to Gerd Moellmann)
2662 ** CALL-NEXT-METHOD with no arguments now passes the original
2663 values of the arguments, even in the presence of assignment.
2664 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2666 ** DISASSEMBLE works with closures and funcallable instances.
2667 ** ADD-METHOD now returns the generic function, not the new method.
2668 ** FIND-METHOD signals an error if the lengths of the specializers
2669 is incompatible with the generic function, even if the ERRORP
2671 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2672 which its argument is a member.
2673 ** DEFCLASS only redefines the class named by its class-name
2674 argument if that name is the proper name of the class;
2675 otherwise, it creates a new class.
2676 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2677 of the UNBOUND-SLOT condition to the name of the slot.
2678 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2679 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2680 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2681 SLOT-MAKUNBOUND in the specified fashion.
2683 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2684 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2685 cross-compilation host. As a consequence, we can now bootstrap our
2686 way up to SBCL starting with a bare gcc toolchain and human-readable
2687 source code (first the source to CLISP, then the source to SBCL).
2688 * A contributed module containing a partial implementation of the
2689 simple-streams interface has been included. (thanks to Rudi
2691 * A contributed module implementing the RFC1321 Message Digest
2692 Algorithm, known as MD5, has been included.
2693 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2694 no longer has any effect, as the code controlled by this feature
2695 has been deleted. (As far as we know, no-one has ever built using
2696 this feature, and its semantics were confused in any case).
2697 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2698 work on conditions (as required by the ANSI specification),
2699 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2700 expected behaviour on conditions. Users should note, however,
2701 that such behaviour is not required by the ANSI specification,
2702 and so use of this behaviour may render their code unportable.
2703 * fixed some bugs revealed by Paul Dietz' test suite:
2704 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2706 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2707 ** SXHASH on condition objects no longer returns NIL.
2708 ** :ALLOCATION :CLASS slots are better treated; their values are
2709 updated on class redefinition, and initforms inherited from
2710 superclasses are applied.
2711 ** REMOVE-METHOD returns its generic function argument even when
2712 no method was removed.
2713 ** SHARED-INITIALIZE now initializes the values of the requested
2714 slots, including those with :ALLOCATION :CLASS.
2715 ** ALLOCATE-INSTANCE now works on structure classes defined via
2716 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2718 ** SLOT-EXISTS-P now works on conditions, as well as structures
2720 ** MAKE-LOAD-FORM now has the required methods on
2721 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2722 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2723 keyword, and now implements the SLOT-NAMES argument.
2724 ** methods with &OPTIONAL arguments no longer allow too many
2725 arguments to be passed in the call without error.
2726 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2727 option is consistent with the required arguments of the generic
2728 function lambda list.
2729 * bug fix: REQUIRE accepts a string designator. (Thanks to
2731 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2732 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2733 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2734 * bug fix: The compiler now checks for duplicated variables in macro
2736 * bug fix: SETQ on globals returns the correct value.
2737 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2738 not a COMPILER-ERROR (followed by some other strange error on
2739 choosing the CONTINUE restart).
2740 * bug fix: make.sh and friends are now more consistent in the way that
2741 they look for GNU "make".
2743 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2744 * experimental native threads support (on x86 Linux >=2.4 only).
2745 This is not compiled in by default: you need to add :SB-THREAD to
2746 the target features. See the "Beyond ANSI" chapter of the manual
2748 * fix for longstanding nonANSIism: The old distinction between
2749 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2750 The return value from CL:FIND-CLASS is now a CLOS class, and
2751 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2752 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2753 * An interface to the MetaObject Protocol, as described in Kiczales,
2754 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2755 MIT Press, 1991, is available from the SB-MOP package.
2756 * incompatible change: the SB-PCL package should now be considered
2757 a private implementation detail, and no longer a semi-private MOP
2759 * minor incompatible change: due to rearrangement for threads, the
2760 control stack and binding stack are now allocated at arbitrary
2761 addresses instead of being hardcoded per-port. Users affected by
2762 this probably have to be doing advanced things with shared
2763 libraries, and will know who they are.
2764 * minor incompatible change: Previously, all --eval forms used were
2765 processed with READ before any of them were processed with EVAL.
2766 Now each --eval form is processed with both READ and EVAL before
2767 the next --eval form is processed. (Thus package operations like
2768 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2769 work as the user might reasonably expect.)
2770 * minor incompatible change: *STANDARD-INPUT* is now only an
2771 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2773 * minor incompatible change: Y-OR-N-P is now character-oriented,
2774 not line oriented. Also, YES-OR-NO-P now works without errors.
2775 (thanks to Antonio Martinez)
2776 * sb-aclrepl module improvements: an integrated inspector, added
2777 repl features, and a bug fix to :trace command.
2778 * Known functions, which cannot be open coded by the backend, are
2779 considered to be able to check types of their arguments. (fixing
2780 a bug report by Nathan J. Froyd)
2781 * fixed a bug in computing method discriminating functions: It is
2782 now possible to define methods specialized on classes which have
2783 forward-referenced superclasses. (thanks to Gerd Moellmann)
2784 * fixed evaluation order in optional entries (reported by Gilbert
2786 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2787 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2788 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2789 required-by-AMOP NAME argument, as well as CLASS and
2790 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2791 * fixed bug 20: DEFMETHOD can define methods using names that are
2792 not the proper names of classes to designate class specializers.
2793 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2794 Unix file descriptors, instead of blowing up. (thanks to Antonio
2796 * Garbage collection refactoring: user-visible change is that a
2797 call to the GC function during WITHOUT-GCING will not do garbage
2798 collection until the end of the WITHOUT-GCING. If you were doing
2799 this you were probably losing anyway.
2800 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2801 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2802 TYPEP the latter but not the former.
2803 * The compiler issues a full WARNING for calls to undefined functions
2804 with names from the CL package.
2805 * MAP-INTO for a vector destination is open coded. (reported by
2806 Brian Downing on c.l.l)
2807 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2808 documentation string.
2809 * fixed some bugs revealed by Paul Dietz' test suite:
2810 ** COPY-ALIST now signals an error if its argument is a dotted
2812 ** Condition slots are now accessed more correctly in the presence
2813 of multiple initargs for a given slot.
2814 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2815 exclude restarts of the same name associated with a different
2817 ** DEFCLASS of forward-referenced classes with another
2818 forward-referenced class in the superclasses list no longer
2820 ** Condition slots are now initialized once each, not multiple
2821 times. (thanks to Gerd Moellmann)
2822 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2824 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2825 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2826 arguments contain duplicated elements.
2827 ** RESTART-CASE understands local macros.
2828 ** RESTART-CASE associates exactly its own restarts with a condition.
2829 ** ENDP in safe mode checks its argument to be of type LIST.
2830 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2831 in question is unbound.
2832 ** Optimized MAKE-INSTANCE functions no longer cause internal
2833 assertion failures in the presence of duplicate initargs.
2834 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2835 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2836 propagated even in the ctor optimized implementation of
2838 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2840 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2841 * a better implementation of SXHASH on (simple) bit vectors,
2842 measured both in execution speed and in distribution of results
2843 over the positive fixnums, has been installed. Likewise, a better
2844 implementation of EQUAL for simple bit vectors is now available.
2845 * fixed CEILING optimization for a divisor of form 2^k.
2846 * fixed bug 240 (emitting extra style warnings "using the lexical
2847 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2848 by Antonio Martinez)
2849 * fixed SXHASH, giving different results for NIL depending on type
2850 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2851 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2852 variable. (found by Rolf Wester)
2853 * SBCL does not ignore type declarations for special
2854 variables. (reported by rif on c.l.l 2003-03-05)
2855 * some bug fixes in contrib/sb-aclrepl/
2856 * fixed some bugs revealed by Paul Dietz' test suite:
2857 ** a bug in the CONS type specifier, whereby the CAR and CDR
2858 types got intertwined, has been fixed;
2859 ** the type system is now able to reason about the interaction
2860 between INTEGER and RATIO types more completely;
2861 ** APPEND, [N]REVERSE and NRECONC check that those their
2862 arguments, which must be proper lists, are really so;
2863 ** An array specialized to be unable to hold elements has been
2864 implemented, as required -- yes, really -- by ANSI;
2865 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2866 on malformed property lists;
2868 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2869 * incompatible packaging change: in line with Unix convention,
2870 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2871 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2872 * REQUIRE and PROVIDE are now optionally capable of doing something
2873 useful. See the documentation string for REQUIRE.
2874 * infrastructure for a managed SBCL contrib system: contributed
2875 modules in this release include:
2876 ** the ASDF system definition facility;
2877 ** an interface to the BSD Sockets API;
2878 ** an ACL-like convenience interface to the repl;
2879 (thanks to Kevin Rosenberg)
2880 ** an implementation of ROTATE-BYTE, with efficient implementation
2882 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2883 gives a closer approximation to the right answer than previously.
2884 (thanks to Raymond Toy)
2885 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2886 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2887 optional environment arguments, as required by ANSI.
2888 * fixed bugs in other functions taking environment objects, allowing
2889 calls with an explicit NIL environment argument to be compiled
2891 * fixed bug 228: primary return values from
2892 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2893 COMPILE or FUNCTION.
2894 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2895 :TYPE will no longer signal errors on innocuous objects.
2896 * fixed bug 231b: SETQ is better at respecting type declarations in
2897 the lexical environment.
2898 * fixed a bug in DEFCLASS: classes named by symbols with no or
2899 unprintable packages can now be defined.
2900 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2901 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2902 * fixed compiler failure related to checking types of functions.
2903 (reported by Robert E. Brown)
2904 * the compiler is now much more consistent in its error-checking
2905 treatment of bounding index arguments to sequence functions: in
2906 (SAFETY 3) code, errors will be signalled in almost all cases if
2907 invalid sequence bounding indices are passed to functions defined
2908 by ANSI to operate on sequences.
2909 * fixed a bug in the build procedure: documentation of SBCL-specific
2910 packages is now preserved and available in the final Lisp image.
2911 * lifted FDEFINITION lookup out of loops in the implementation of
2912 many list operations. (thanks to Robert E. Brown)
2913 * fixed a bug in the reader: the #n# reader macro now works for
2914 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2915 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2916 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2917 E. Brown and Tony Martinez respectively)
2918 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2919 SLOT-BOUNDP: in particular, it is now able to optimize them much
2920 better, and is now not vulnerable to having packages renamed.
2921 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2922 always called when a slot is not present in an instance. (thanks
2924 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2925 longer signal an error on unreading a character following EOF on
2926 the previous constituent stream. (thanks to Tony Martinez)
2927 * fixed some bugs revealed by Paul Dietz' test suite:
2928 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2929 not just nonnegative fixnums;
2930 ** the logical bit-array operators such as BIT-AND now accept an
2931 explicit NIL for their "opt-arg" argument (to indicate a
2932 freshly-consed result bit-array);
2933 ** ELT now signals an error on an invalid sequence index in safe
2935 ** the type system is now cleverer about negations of numeric
2936 types, and consequently understands the BIGNUM and RATIO types
2938 ** the type system is now cleverer about the interaction between
2939 INTEGER and RATIO types: while bugs still remain, many more
2940 cases are accurately computed;
2941 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2942 if it is in the last clause;
2943 ** CONSTANTLY now correctly returns a side-effect-free function in
2945 ** DECLARE is no longer treated as a special-operator; in
2946 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2947 * incremented fasl file version number due to the change in the
2948 DEFSTRUCT-SLOT-DESCRIPTION structure.
2950 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2951 * minor incompatible change: code processed by the "interpreter" or
2952 EVAL now has a compilation optimization policy of (DEBUG 2)
2953 (changed from (DEBUG 1)) to improve debuggability of interactive
2954 development, and to allow the use of the debug RETURN command in
2956 * an experimental implementation of the RETURN command for the
2957 debugger has been included. (thanks to Frederik Kuivinen)
2958 * fixed bug 62: constraints were not propagated into a loop.
2959 * fixed bug in embedded calls of SORT (reported and investigated by
2961 * fixed some bugs revealed by Paul Dietz' test suite:
2962 ** printing and reading of arrays with some dimensions having
2963 length 0 (thanks to Gerd Moellmann);
2964 ** BOA constructor with &AUX argument without a default value does
2965 not cause a type error;
2966 ** CONSTANTP now returns true for all self-evaluating objects.
2968 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2969 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2970 accessors that are related by inheritance, as specified in the
2971 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2972 Valtteri Vuorikoski)
2973 * The compiler is now able to inline functions that were defined in
2974 a complex lexical environment (e.g. inside a MACROLET).
2975 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2977 * Support for the upcoming FreeBSD-5.0 release has been included.
2978 (thanks to Dag-Erling Smorgrav)
2979 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2980 effect when it is not in a toplevel context.
2981 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2982 stand a better chance of being correct. (thanks to Gerd
2984 * fixed bug in COERCE, which now signals an error on coercing a
2985 rational to a bounded real type which excludes the expected
2987 * The compiler is now able to derive types more accurately from the
2988 COERCE and COMPILE functions.
2989 * fixed bug 223: functional binding is considered to be constant
2990 only for symbols in the CL package.
2991 * fixed bug 231: SETQ did not check the type of a variable being set
2992 (reported by Robert E. Brown)
2993 * A new optimization for MAKE-INSTANCE has been included, fixing
2994 various bugs (including relating to :ALLOCATION :CLASS slots and
2995 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2997 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2998 ** As required by ANSI, LOOP now disallows anonymous collection
2999 clauses such as COLLECT I in conjunction with aggregate boolean
3000 clauses such as THEREIS (= I 1);
3001 ** LOOP now signals an error when any variable is reused in the
3002 same loop (including the potentially useful construct analogous
3003 to WITH A = 1 WITH A = (1+ A);
3004 ** IT is only a special loop symbol within the first clause of a
3005 conditional loop clause;
3006 ** LOOP with a typed iteration variable over a hashtable now
3007 signals a type error iff it should.
3008 * fixed some other bugs revealed by Paul Dietz' test suite:
3009 ** FILE-STREAM now names the class previously known as FD-STREAM;
3010 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3011 argument) no longer signals an error;
3012 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3013 of the slot symbol, rather than using the current package
3014 ((:CONC-NAME "") continues to intern the slot's name in the
3016 * incremented fasl file version number, because of the incompatible
3017 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3018 of the new implementation of DEFINE-COMPILER-MACRO.
3020 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3021 * Support for building SBCL for MIPS platforms running in
3022 little-endian mode has now been checked in, and basic
3023 functionality on said platforms verified.
3024 * minor incompatible change: PCL now records the pathname of a file
3025 in which methods and the like are defined, rather than its
3027 * minor incompatible change: TRUENAME now considers the truename of
3028 a file naming a directory to be the pathname with :DIRECTORY
3029 component indicating that directory.
3030 * minor incompatible change: a NAMED clause in the extended form of
3031 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3032 reason for the previous behaviour is unclear.
3033 * more systematization and improvement of CLOS and MOP conformance
3034 in PCL (thanks to Gerd Moellman and Pierre Mai):
3035 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3037 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3038 primary methods with no specializers;
3039 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3041 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3042 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3043 FORWARD-REFERENCED-CLASSes; error reporting on
3044 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3046 ** SXHASH on CLOS instances now uses a slot internal to the
3047 instance to return different numbers on distinct instances,
3048 while preserving the same return value through invocations of
3050 ** DEFMETHOD signals errors when methods with longer incongruent
3051 lambda lists are added to generic functions;
3052 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3053 CLASS, as specified in AMOP;
3054 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3055 on the order returned by the primary method for classes of
3056 class STANDARD-CLASS;
3057 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3058 * fixed some bugs shown by Paul Dietz' test suite:
3059 ** DOLIST puts its body in TAGBODY;
3060 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3062 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3063 value producing form;
3064 ** if more variables are given to PROGV than values, extra
3065 variables are bound and made to have no value;
3066 ** NSUBSTITUTE on list arguments gets the right answer with
3068 ** ELT signals an error of type TYPE-ERROR when the index argument
3069 is not a valid sequence index;
3070 ** LOOP signals (at macroexpansion time) an error of type
3071 PROGRAM-ERROR when duplicate variable names are found;
3072 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3073 ** FUNCALL of special-operators now cause an error of type
3075 ** PSETQ now works as required in the presence of side-effecting
3076 symbol-macro places;
3077 ** NCONC accepts any object as its last argument;
3078 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3080 ** loop-for-as-package does not require a package to be explicitely
3082 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3084 * fixed bug 166: compiler preserves "there is a way to go"
3085 invariant when deleting code.
3086 * fixed bug 172: macro lambda lists with required arguments after
3087 &REST arguments now cause an error to be signalled. (thanks to
3089 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3090 bug. (thanks to Matthew Danish)
3091 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3093 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3094 arguments are lexically rebound. (thanks to Gerd Moellmann and
3096 * fixed bug 194: error messages are now more informative when there
3097 is no primary method applicable in a call to a generic
3098 function. (thanks to Gerd Moellmann)
3099 * fixed bug in command line argument checking (thanks to Julian
3101 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3102 (thanks to Matthew Danish)
3103 * incremented fasl file version number, because of the
3104 SXHASH-related changes in the layout of CLOS data structures
3106 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3107 * minor incompatible change: The runtime (the Unix executable named
3108 "sbcl") is now much pickier about the .core files it will load.
3109 Essentially it now requires .core files to descend from the same
3110 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3111 as the runtime does. (The intent is to prevent the crashes which
3112 can occur, and which can even be reported as mysterious failures,
3113 when people patch the sources or change the build parameters
3114 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3115 sbcl and .core files.)
3116 * fixed bug: VALUES-LIST is no longer optimized away.
3117 * fixed bug 142: The FFI conversion of C string values to Lisp
3118 string values no longer conses excessively. (thanks to Nathan
3119 Froyd porting Raymond Toy's fix to CMU CL)
3120 * began to systematize and improve MOP conformance in PCL (thanks to
3121 Nathan Froyd, Gerd Moellman and Pierre Mai):
3122 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3124 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3125 ** FINALIZE-INHERITANCE is now called on class finalization;
3126 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3127 argument precedence order.
3128 * fixed bug 202: The compiler no longer fails on functions whose
3129 derived types contradict their declared type.
3130 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3131 so it can be non-toplevel.
3132 * The fasl file version number has changed (because of the new
3133 implementation of DEFMACRO).
3134 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3135 safe code, that any length requirement by their type-specifier
3136 argument is valid. The exceptions to this are described in bug
3138 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3139 functions now check the types of their inputs as required by ANSI.
3140 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3141 introduced symbol is DECLAREd to be SPECIAL.
3142 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3143 * fixed bug: :COUNT argument to sequence functions may be negative
3144 * fixed bug: body of DO-SYMBOLS may contain declarations
3145 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3146 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3147 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3148 symbol macro only once
3149 * fixed printing of call frame when argument list is unavailable
3150 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3151 * compiler no longer signals WARNING on unknown keyword
3154 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3155 * A beta-quality port to the mips architecture running Linux,
3156 based on the old CMUCL backend, has been made. It has been tested
3157 on a big-endian kernel, and works sufficiently well to be able to
3158 rebuild itself; it has not been tested in little-endian mode.
3159 * fixed an inconsistency between gencgc.c and purify.c which made
3160 dumping/loading .core files unreliable
3161 * fixed bug 120a: The compiler now deals correctly with IFs where
3162 the consequent is the same as the alternative, instead of
3163 misderiving the return type. (thanks to Alexey Dejneka)
3164 * fixed bug 113: Logical pathnames are now dumpable (the logical
3165 host is resolved at load-time, throwing an error if it is not
3167 * fixed bug 174: FORMAT's error message is slightly clearer when a
3168 non-printing character is used in a format directive.
3169 * fixed several bugs in compiler checking of type declarations, i.e.
3170 violations of the Python "declarations are assertions" principle
3171 (thanks to Alexey Dejneka)
3172 * fixed several bugs in PCL's error checking (thanks to Gerd
3174 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3176 * fixed bug in compilation of functions as first class values
3177 (thanks to Antonio Martinez-Shotton)
3178 * The compiler's handling TYPE-ERRORs which it can prove will
3179 inevitably happen at runtime has been cleaned up and corrected
3180 in several ways. (thanks to Alexey Dejneka)
3181 * improved argument type checking for various basic arithmetic
3182 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3183 lavished on them in the past that they can be compiled in many
3184 ways in different special cases
3185 * fixed bug 181: compiler checks validity of user supplied type
3187 * cleaned up code flushing in optimization: Function calls which
3188 should signal errors for safety purposes (e.g. which ANSI says
3189 should signal errors when their arguments are of incorrect type)
3190 are no longer optimized away.
3191 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3192 * incremented fasl file version number, because changes in the
3193 implementation of sequence functions like COERCE caused
3194 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3195 in old inline expansions) to become undefined. (Actually these
3196 changes were later undone, so we might very well be binary
3197 compatible with 0.7.7 after all, but leaving the version number
3198 incremented seemed like the simplest and most conservative
3201 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3202 * An alpha-quality port to the parisc architecture running Linux,
3203 based on the old CMUCL backend, has been made. This, even more so
3204 than the other backends, should be considered still a work in
3205 progress; known problems include that the Linux kernel in 64-bit
3206 mode does not propagate the correct sigcontext structure to
3207 userspace, and consequently SBCL on a parisc64 kernel will not
3209 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3210 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3211 suppressed.) Also now that the compiler is looking at declarations
3212 in the environment, it checks optimization declarations as well,
3213 and suppresses inlining when (> DEBUG SPEED).
3214 * More fixes have been made to treatment of floating point exception
3215 treatment and other Unix signals. In particular, floating point
3216 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3217 * The detection and handling of control stack exhaustion (infinite
3218 or very deeply nested recursion) has changed. Stack exhaustion
3219 detection is now done by write-protecting pages at the OS level
3220 and applies at all optimization settings; when found, a
3221 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3222 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3223 be caught using IGNORE-ERRORS.
3224 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3225 correctly when transforming recursive LABELS functions to LETs.
3226 (thanks to Alexey Dejneka)
3227 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3228 that are names of constants or global variables.
3229 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3230 alien routines with docstrings.
3231 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3232 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3234 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3235 (thanks to Eric Marsden)
3236 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3237 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3238 * Bug fix: The SPARC backend can now compile functions involving
3239 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3240 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3241 to a functions expecting a PATHNAME-DESIGNATOR.
3242 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3243 lambda lists. (thanks to Alexey Dejneka)
3244 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3245 functions so that EXPORTing the name of the class doesn't cause
3246 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3247 to Antonio Martinez for reporting this)
3248 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3249 constant-folded without failing an assertion. (thanks to Einar
3250 Floystad Dorum for reporting this)
3251 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3252 types are now dealt with more correctly by the compiler.
3253 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3254 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3255 behaviour analogous to RENAME-FILE. This puts its behaviour more
3256 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3258 * The fasl file version number has changed again. (because of the
3259 bug fix involving the names of PCL MAKE-INSTANCE functions)
3261 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3262 * bug fix: Floating point exceptions are treated much more
3263 consistently on the x86/Linux and PPC/Linux platforms.
3264 * Array initialization with :INITIAL-ELEMENT is now much faster for
3265 cases when the compiler cannot open code the array creation, but
3266 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3267 array accesses have also seen a speed increase.
3268 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3269 specified. (This was at the root of some bad interactions between
3270 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3272 * bug fix: Internal error arguments for undefined functions are now
3273 computed correctly on the PPC/Linux platform.
3274 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3275 Raymond Toy's patch for CMU CL)
3276 * Support for the Solaris 9 operating environment has been included
3277 (thanks to Daniel Merritt)
3278 * A very ugly but hopefully complete draft of the missing FFI chapter
3279 of the manual has been created by reformatting the corresponding
3280 CMU CL manual chapter into (currently very ugly and incoherent)
3281 DocBook and bringing it up to date for SBCL behavior. Thus, the
3282 manual is now essentially complete, at least by my extreme
3283 once-and-only-once standards, whereby it's acceptable to refer to
3284 the doc strings of SB-EXT functions as the primary documentation.
3285 * The fasl file version number has changed again, due to cleanup of
3286 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3287 is no longer a static symbol.)
3289 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3290 * SBCL now builds with OpenMCL (version 0.12) as the
3291 cross-compilation host; also, more progress has been made toward
3292 bootstrapping under CLISP.
3293 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3295 * bug 158 fixed: The compiler can now deal with integer loop
3296 increments different from 1; fixing this turned out also to fix
3298 * bug 169 fixed: no more bogus warnings about using lexical bindings
3299 despite the presence of perfectly good SPECIAL declarations (thanks
3300 to David Lichteblau)
3301 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3302 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3303 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3305 * bug 180 fixed: Method combination specifications no longer ignore
3306 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3307 * bug fix: Structure type predicate functions now check their argument
3308 count as they should.
3309 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3310 correctly. (thanks to Pierre Mai)
3311 * minor incompatible change: The --noprogrammer option is deprecated
3312 in favor of the new --disable-debugger option, which is very similar.
3313 (The major difference is that it takes effect at a slightly different
3314 time at startup, causing handling of errors in --sysinit and
3315 --userinit files will be affected differently.) The
3316 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3317 been added to allow this functionality to be controlled from ordinary
3318 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3319 maintainers, who might want to run non-interactive scripts to
3320 build SBCL cores which will later be used interactively.)
3321 * minor incompatible change: The LOAD function no longer, when given
3322 a wild pathname to load, loads all files matching that pathname.
3323 Instead, an error of type FILE-ERROR is signalled.
3325 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3326 * bug 147 fixed: The compiler preserves its block link/count
3327 invariants more correctly now so that it doesn't crash. (thanks
3329 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3331 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3332 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3333 no-value-for-this-character as NIL (as specified by ANSI).
3334 * HOST-NAMESTRING on physical pathnames now returns a string that is
3335 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3336 (thanks to Christophe Rhodes)
3337 * The Alpha port handles icache flushing more correctly. (thanks to
3339 * More progress has been made toward bootstrapping under CLISP. (thanks
3340 to Christophe Rhodes)
3341 * The fasl file format has changed again, because dynamic loading
3342 on OpenBSD (which has non-ELF object files) motivated some cleanups
3343 in the way that foreign symbols are transformed and passed around.
3344 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3345 is no longer treated as whitespace by the reader, but instead as
3346 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3347 |A
\7fB|, instead of A as it used to.
3349 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3350 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3351 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3352 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3353 this as of 0.7.1.45, but was left out of the previous news section
3354 (thanks to Dan Barlow)
3355 * SBCL now runs on the Solaris operating system on SPARC architectures
3356 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3357 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3358 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3359 * SPARC backend cleanups, allowing builds of cores optimized for V8
3360 and V9 SPARCS, and also emission of code targeted to a particular
3361 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3363 * SBCL is closer to bootstrapping under CLISP, thanks to various
3364 fixes by Christophe Rhodes.
3365 * The fasl file format has changed again, to allow the compiler's
3366 INFO database to support symbol macros.
3367 * The user manual (in doc/) is formatted into HTML more nicely.
3368 (thanks to coreythomas)
3369 * The system is smarter about SUBTYPEP relationships, especially
3370 those involving NOT types (including types such as ATOM which are
3371 represented internally using NOT types). Thus SUBTYPEP is less
3372 likely to return (VALUES NIL NIL) in general, and in particular
3373 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3375 * The fasl file format has changed again, because the internal
3376 representation of types now includes a new slot to support the new
3377 SUBTYPEP-of-NOT-types logic.
3378 * (not a change in the main branch of SBCL, but a related prototype
3379 which can hopefully be merged into the main branch of SBCL in the
3380 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3381 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3382 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3383 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3384 dispatch macro character combinations. (thanks to Alexey Dejneka)
3385 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3386 that superficially look like logical namestrings correctly.
3387 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3389 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3390 default supplied-p) for &optional and &key arguments. (thanks to
3393 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3394 * incompatible change: The compiler is now less aggressive about
3395 tail call optimization, doing it only when (> SPACE DEBUG) or
3396 (> SPEED DEBUG). (This is an incompatible change because there are
3397 programs which relied on the old CMU-CL-style behavior to optimize
3398 away their unbounded recursion which will now die of stack overflow.)
3399 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3400 for non-GENCGC systems has been increased to 20M (since that
3401 seems much closer to the likely performance optimum for modern
3402 systems than the old 4M value was)
3403 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3404 and *DEBUG-PRINT-LEVEL*
3405 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3406 of CMU CL's support for SPARC, and various endianness and other
3407 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3408 * new syntactic sugar for the Unix command line: --load foo.bar is now
3409 an alternate notation for --eval '(load "foo.bar")'.
3411 ** The system now detects stack overflow and handles it gracefully,
3412 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3413 optimization settings. (This is a good thing in general, and
3414 its introduction in this version should be particularly timely
3415 for anyone whose code fails because of suppression of tail
3417 ** The system now hunts for the C variable "environ" in a more
3418 devious way, to avoid segfaults when the C library version differs
3419 between compile time and run time. (thanks to Christophe Rhodes)
3420 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3421 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3422 ** The compiler no longer issues bogus style warnings for undefined
3423 classes in the same source file as the DEFCLASSes which defined
3424 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3426 ** fixes in CONDITION class precedence list for undefined function
3427 errors (thanks to Alexei Dejneka)
3428 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3429 correctly. (thanks to Dan Barlow)
3430 ** portability fixes aiming at bootstrapping under CLISP (thanks
3431 to Dave McDonald and Christophe Rhodes)
3432 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3433 ** fixes in type translation and and type inference (thanks to
3435 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3436 ** various fixes in the new ports (thanks to Dan Barlow)
3437 * several changes related to debugging:
3438 ** suppression of tail recursion, as noted above
3439 ** stack overflow detection, as noted above
3440 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3441 is now the default. (For some time encapsulation has been more
3442 reliable than the breakpoint-based :ENCAPSULATE NIL
3443 implementation, at least on X86 systems; and I just noticed that
3444 encapsulation also seems closer to the spirit of the ANSI
3447 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3449 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3450 up properly again. (There was a packaging bug in 0.7.0 which
3451 left their definitions in SB-SYS::LOAD-FOREIGN and
3452 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3453 vital for most things which interface to C-level interfaces,
3454 like extensions working with sockets or databases or
3455 Perl-compatible regexes or whatever, and the need to fix
3456 this bug is the main reason that 0.7.1 was released so
3458 ** DEFGENERIC is now choosier about the methods it redefines, so that
3459 reLOADing a previously-LOADed file containing DEFGENERICs does
3460 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3461 cycle now works as it should. (thanks to Alexey Dejneka)
3462 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3463 fixed by Christophe Rhodes. (He actually submitted this patch
3464 months ago, and I delayed until after 0.7.0.)
3465 ** Bug 111 (internal compiler confusion about runtime checks on
3466 FUNCTION types) was fixed by Alexey Dejneka.
3467 * Some internal cleanups (getting rid of variables which aren't
3468 needed now that the byte interpreter is gone) caused the fasl
3469 file format number to change again.
3471 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3472 * major incompatible change: The default fasl file extension, i.e. the
3473 default extension for files produced by COMPILE-FILE, has changed
3474 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3476 ** There are many changes in the implementation of the compiler.
3477 SBCL is now essentially a compiler-only implementation of ANSI
3478 Common Lisp. EVAL still "interprets" a few special cases, but
3479 almost all the interesting cases are handled by creating
3480 a LAMBDA expression, calling COMPILE on it, then calling
3481 FUNCALL on the result.
3482 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3483 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3484 Since the code is newer, there might still be some new bugs
3485 (though not as many as before Martin Atzmueller's fixes:-). But
3486 the new code is substantially simpler and clearer, and hopefully
3487 any remaining bugs will be simpler, less fundamental, and more
3488 fixable then the bugs in the old code.
3489 ** The revised compiler is still a little unsteady on its feet.
3491 *** The debugging information it produces (particularly the names
3492 of FUNCTION objects) is sometimes much less useful than what
3493 the old compiler produced.
3494 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3495 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3496 within a MACROLET) has been temporarily weakened.
3497 ** There are new compiler optimizations for various functions:
3498 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3499 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3500 *** the math functions TRUNCATE, FLOOR, and CEILING
3501 *** the function-of-all-trades COERCE
3502 Mostly these should be transparent, but there's one
3503 potentially-annoying problem (bug 117): when the compiler
3504 inline-expands a function and does type analysis on the result,
3505 it can create control paths which have type mismatches, and
3506 when it can't prove that those control paths aren't taken,
3507 it will issue WARNINGs about the type mismatches. This is
3508 a particular problem in practice for the new sequence functions.
3509 It's not clear how this should be fixed, and for now, a
3510 workaround is given in the entry for 117 in the BUGS file.
3511 ** (Because of the interaction between the two previous items --
3512 occasional inlining problems and new inline expansions -- some
3513 of the new sequence function optimizations won't really kick in
3514 completely until debugging information, and then inlining, are
3515 straightened out in some future version.)
3516 * minor incompatible changes:
3517 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3518 DIRECTORY no longer implicitly promotes NIL slots of its
3519 pathname argument to :WILD. In particular, when you ask for the
3520 contents of a directory (which you used to be able to do without
3521 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3522 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3523 ** changes in behavior that ANSI explicitly defines to be
3524 implementation dependent:
3525 *** The new compiler-only implementation still conforms with ANSI,
3526 but acts a little different than before. Besides the obvious
3527 changes in performance tradeoffs (that the cost per form passed
3528 to EVAL has gone up, and the cost per form executed by EVAL
3529 has gone down), the behavior of the system changes a little
3530 because there are no longer any interpreted function objects.
3531 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3532 e.g. doing COMPILE on the output of interactive DEFUN is
3534 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3536 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3537 has changed from (:CL) to NIL.
3538 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3539 e.g. #\Newline and #\Tab, don't have names specified in the
3540 ANSI Common Lisp standard, is now based on their ASCII symbolic
3541 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3542 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3543 are no longer used for output.
3544 ** changes in internal implementation constants:
3545 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3546 4 million. (If your application spends a lot of time GCing and
3547 you have a lot of RAM, you might want to experiment with
3548 increasing it even more.)
3549 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3550 However, almost all old code should still continue to work without
3551 immediate update, as SB-C-CALL is now a (deprecated) nickname
3553 ** Old operator names in the style DEF-FOO are now deprecated in
3554 favor of new corresponding names DEFINE-FOO, for consistency with
3555 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3556 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3557 internal symbols, but a few supported extensions like
3558 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3559 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3560 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3561 etc. as you get deeper into recursive calls to the debugger
3562 command loop, instead of the old "5]", "5]]", "5]]]"
3563 sequence. (I was motivated to do this when squabbles between
3564 ILISP and SBCL left me very deeply nested in the debugger. In the
3565 short term, this change will probably provoke more ILISP/SBCL
3566 squabbles, but hopefully it will be an improvement in the long run.)
3567 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3568 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3569 function, since declaiming return types involving aliens is
3570 (1) annoyingly messy to do by hand and (2) vital to efficient
3571 compilation of code which calls such functions.
3572 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3573 longer reexported by the SB-EXT package. They're solely useful
3574 for alien code, so it seems more logical that you should get
3575 them from the SB-ALIEN package, not in SB-EXT.
3576 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3577 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3578 features. Instead, the code that they used to control is always
3579 built into the system.
3580 * many other bug fixes
3581 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3582 advantage of the new EVAL-WHEN stuff and to clean them up in
3583 general, and they are now more ANSI-compliant in a number of
3584 ways. Martin Atzmueller is responsible for a lot of this.
3585 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3587 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3588 *** correct ERROR type for various file operations
3589 *** some fixes for Lisp streams
3590 *** DEFMETHOD syntax checking
3591 *** changing old weird representation of debug information as
3592 strings (which, among their other deficiencies, don't transform
3593 correctly when you rename packages, and don't change their
3594 print representation when you change things like *PACKAGE*
3595 and *PRINT-LENGTH*) to symbols and lists of symbols
3596 He also made several improvements and fixed several bugs in DESCRIBE.
3597 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3599 *** misbehavior of WRITE-STRING/WRITE-LINE
3600 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3601 and several other LOOP problems as well
3602 *** DIRECTORY when similar filenames are present
3603 *** DEFGENERIC with :METHOD options
3604 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3605 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3606 *** argument ordering in FIND with :TEST option
3607 *** mishandled package designator argument in APROPOS-LIST
3608 *** various problems in the backquote readmacro
3609 *** a bug in APROPOS
3610 *** probably some others that I'm not describing very well here,
3611 since the CVS log documents them by reference to sbcl-devel
3612 messages, and the SourceForge archives aren't working well.:-(
3613 ** Dan Barlow improved the Alpha port (and is making progress on the
3614 PPC port, for those of you who think different).
3615 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3616 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3617 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3618 continues to work on the SPARC port (for those of you in a position
3619 to look down upon our little PC-compatible boxes from a great height).
3620 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3621 and uses that copy, rather than the current dynamic value, when
3622 it's trying to decide whether to truncate output. Thus e.g.
3623 (let ((*print-lines* 50))
3624 (pprint-logical-block (stream nil)
3626 (let ((*print-lines* 8))
3627 (print (aref possiblybigthings i) stream)))))
3628 should now truncate the logical block only at 50 lines, instead of
3629 often truncating it at 8 lines, as it did before.
3630 * The doc/cmucl/ directory, containing old CMU CL documentation
3631 from the time of the fork, is no longer part of the base system.
3632 SourceForge has shut down its anonymous FTP service, and with it
3633 my original plan for distributing the old CMU CL documentation
3634 there. For now, if you need these files you can download an old
3635 SBCL source release and extract them from it.
3636 * The fasl file version number changed again, for dozens of reasons,
3637 some of which are apparent above.
3639 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3640 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3641 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3642 patch, so that DISASSEMBLE output is much nicer.
3643 * The code in the SB-PROFILE package now seems reasonably stable.
3644 I still haven't decided what the final interface should look like
3645 (I'd like PROFILE to interact cleanly with TRACE, since both
3646 facilities use function encapsulation) but if you have a need
3647 for profiling now, you can probably use it successfully with
3648 the current CMU-CL-style interface.
3649 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3650 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3651 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3652 link now returns the dangling link itself, and for similar
3653 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3654 link itself. (In these cases the old code signalled an error and
3655 looped endlessly, respectively.) Thus, DIRECTORY now works even
3656 in the presence of dangling and cyclic symbolic links.
3657 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3658 is now a supported extension again, since the consensus on
3659 sbcl-devel was that it can be useful for ordinary development
3660 work, not just for debugging SBCL itself.
3661 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3662 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3663 that currently-defined functions might be redefined later with
3664 different return types.
3665 * Hash tables can be printed readably, as inspired by CMU CL code
3666 of Eric Marsden and SBCL code of Martin Atzmueller.
3667 * better error handling in CLOS method combination, thanks to
3668 Martin Atzmueller porting Pierre Mai's CMU CL patches
3669 * more overflow fixes for >16Mbyte I/O buffers
3670 * A bug in READ has been fixed, so that now a single Ctrl-D
3671 character suffices to cause end-of-file on character streams.
3672 In particular, now you only need one Ctrl-D at the command
3673 line (not two) to exit SBCL.
3674 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3676 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3677 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3678 does the right thing.
3679 * The implementation of some type tests, especially for CONDITION
3680 types, is now tidier and maybe faster, due to CMU CL code
3681 originally by Douglas Crosher, ported by Martin Atzmueller.
3682 * Some math functions have been fixed, and there are new
3683 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3684 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3685 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3686 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3687 likely to be built into sbcl-0.7.0.)
3688 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3689 CL, ported to SBCL by Martin Atzmueller)
3690 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3691 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3692 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3693 * There's a new slam.sh hack to shorten the edit/compile/debug
3694 cycle for low-level changes to SBCL itself, and a new
3695 :SB-AFTER-XC-CORE target feature to control the generation of
3696 the after-xc.core file needed by slam.sh.
3697 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3698 COMPILE-FILE is no longer supported, so that now every function
3699 gets an entry point, so that block compilation looks a little
3700 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3701 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3702 returns the number of bytes consed since the system started,
3703 rather than the number consed since the first time the function
3704 was called. (The new definition parallels ANSI functions like
3705 CL:GET-INTERNAL-RUN-TIME.)
3706 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3707 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3708 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3709 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3710 * Fasl file version numbers are now independent of the target CPU,
3711 since historically most system changes which required version
3712 number changes have affected all CPUs equally. Similarly,
3713 the byte fasl file version is now equal to the ordinary
3716 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3717 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3718 is no longer recognized. I apologize for this, because it was
3719 listed in SB-EXT as a supported extension, but I found that
3720 its existing behavior was poorly specified, as well as incorrectly
3721 specified, and it looked like too much of a mess to straighten it
3722 out. I have enough on my hands trying to get ANSI stuff to work..
3723 * many patches ported from CMU CL by Martin Atzmueller, with
3724 half a dozen bug fixes in pretty-printing and the debugger, and
3725 half a dozen others elsewhere
3726 * fixed bug 13: Floating point infinities are now supported again.
3727 They might still be a little bit flaky, but thanks to bug reports
3728 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3729 as flaky as they were.
3730 * The --noprogrammer command line option is now supported. (Its
3731 behavior is slightly different in detail from what the old man
3732 page claimed it would do, but it's still appropriate under the
3733 same circumstances that the man page talks about.)
3734 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3735 are now supported, and enabled by default. Thus, the compiler can
3736 handle many floating point and complex operations much less
3737 inefficiently. (Thus e.g. you can implement a complex FFT
3739 * The compiler now detects type mismatches between DECLAIM FTYPE
3740 and DEFUN better, and implements CHECK-TYPE more correctly, and
3741 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3742 thanks to patches from Martin Atzmueller.
3743 * various fixes to make the cross-compiler more portable to
3744 ANSI-conforming-but-different cross-compilation hosts (notably
3745 Lispworks for Windows, following bug reports from Arthur Lemmens)
3746 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3747 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3748 fixed thanks to Pierre Mai's CMU CL patches.
3749 * fixes to keep the system from overflowing internal counters when
3750 it tries to use i/o buffers larger than 16M bytes
3751 * fixed bug 45a: Various internal functions required to support
3752 complex special functions have been merged from CMU CL sources.
3753 (When I was first setting up SBCL, I misunderstood a compile-time
3754 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3755 * improved support for type intersection and union, fixing bug 12
3756 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3757 more obscure bugs as well
3758 * some steps toward byte-compiling non-performance-critical
3759 parts of the system, courtesy of patches from Martin Atzmueller
3760 * Christophe Rhodes has made some debian packages of sbcl at
3761 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3762 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3763 stable, but are nonetheless usable. When he's ready, I'd be happy
3764 to add them to the SourceForge "File Releases" section. (And if
3765 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3766 * new fasl file format version number (because of changes in
3767 internal representation of (OR ..) types to accommodate the new
3768 support for (AND ..) types, among other things)
3770 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3771 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3773 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3775 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3776 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3777 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3778 * DESCRIBE now works on structure objects again.
3779 * Most function call argument type mismatches are now handled as
3780 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3781 know whether the function will be redefined before the call is
3782 executed. (The compiler could flag local calls with full WARNINGs,
3783 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3784 it doesn't keep track of enough information to know whether calls
3785 are local in this sense.)
3786 * Compiler output is now more verbose, with messages truncated
3787 later than before. (There should be some supported way for users
3788 to override the default verbosity, but I haven't decided how to
3789 provide it yet, so this behavior is still controlled by the internal
3790 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3791 src/compiler/ir1util.lisp.)
3792 * Fasl file format version numbers have increased again, because
3793 support for the Gray streams extension changes the layout of the
3794 system's STREAM objects.
3795 * The Gray subclassable streams extension now works, thanks to a
3796 patch from Martin Atzmueller.
3797 * The full LOAD-FOREIGN extension (not just the primitive
3798 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3799 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3800 but like most other programs, it defaults to copying the Unix
3801 environment from the original process instead of starting the
3802 new process in an empty environment.
3803 * Extensions which manipulate the Unix environment now support
3804 an :ENVIRONMENT keyword option which doesn't smash case or
3805 do other bad things. The CMU-CL-style :ENV option is retained
3806 for porting convenience.
3807 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3808 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3810 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3812 * A patch from Martin Atzmueller seems to have solved the SIGINT
3813 problem, and as far as we know, signal-handling now works cleanly.
3814 (If you find any new bugs, please report them!)
3815 * The system no longer defaults Lisp source file names to types
3816 ".l", ".cl", or ".lsp", but only to ".lisp".
3817 * The compiler no longer uses special default file extensions for
3818 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3819 seems to expect a single default extension for all compiled code,
3820 and there's no compelling reason to try to stretch the standard
3821 to allow two different extensions.) Instead, byte-compiled files
3822 default to the same extension as native-compiled files.
3823 * Fasl file format version numbers have increased again, because
3824 a rearrangement of internal implementation packages made some
3825 dumped symbols in old fasl files unreadable in new cores.
3826 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3827 many fewer weird special cases.
3828 * Bug #17 (differing COMPILE-FILE behavior between logical and
3829 physical pathnames) has been fixed, and some related misbehavior too,
3830 thanks to a patch from Martin Atzmueller.
3831 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3832 by Tim Moore, ported to SBCL by Martin Atzmueller.
3833 * Martin Atzmueller fixed several filesystem-related problems,
3834 including bug #36, in part by porting CMU CL patches, which were
3835 written in part by Paul Werkowski.
3836 * More compiler warnings in src/runtime/ are gone, thanks to
3837 more patches from Martin Atzmueller.
3838 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3841 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3843 * DESCRIBE now works on CONDITION objects.
3844 * The debugger now handles errors which arise when trying to print
3845 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3847 * The build system now uses an additional file, customize-target-features.lisp,
3848 to allow local modifications to the target *FEATURES* list. (The point of
3849 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3850 debugging features enabled, without having to worry about propagating it
3851 into everyone's system when I do a "cvs update".) When no
3852 customize-target-features.lisp file exists, the target *FEATURES* list
3853 should be constructed the same way as before.
3854 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3855 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3856 warn about behavior which is undefined under the ANSI spec. Note: This
3857 is specified by ANSI, but it's not very popular with programmers.
3858 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3859 macro in the SBCL sources for an example of a workaround which you
3860 might use to make portable ANSI-standard code which does what you want.
3861 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3862 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3863 when it's used as a non-toplevel form. (This is required in order
3864 to implement the DEFCONSTANT-EQX macro.)
3865 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3866 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3867 funny things, due to bugs in the implementation of EVAL-WHEN
3868 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3869 non-toplevel DEFCONSTANTs are uncommon.)
3870 * The core file version number and fasl file version number have been
3871 incremented, because the old noncompliant DEFCONSTANT behavior involved
3872 calling functions which no longer exist, and because I also took the
3873 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3874 * fixed bug 1 (error handling before read-eval-print loop starts), and
3875 redid debugger restarts and related debugger commands somewhat while
3877 ** The QUIT debugger command is gone, since it did something
3878 rather different than the SB-EXT:QUIT command, and since it never
3879 worked properly outside the main toplevel read/eval/print loop.
3880 Invoking the new TOPLEVEL restart provides the same functionality.
3881 ** The GO debugger command is also gone, since you can just invoke
3882 the CONTINUE restart directly instead.
3883 ** The TOP debugger command is also gone, since it's redundant with the
3884 FRAME 0 command, and since it interfered with abbreviations for the
3886 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3888 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3889 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3890 some porting work by Martin Atzmueller)
3891 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3892 with a boatload of patches which he ported to SBCL. Now that those
3894 ** The system tries to make sure that its low-priority messages
3895 are prefixed by semicolons, to help people who like to use
3896 syntax highlighting in their ILISP buffer. (This patch
3897 was originally due to Raymond Toy.)
3898 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3899 patches originally written by Raymond Toy.
3900 ** The compiler understands coercion between single-value and
3901 multiple-VALUES type expressions better, getting rid of some very
3902 weird behavior, thanks to patches originally by Robert MacLachlan
3903 and Douglas Crosher.
3904 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3905 lambda lists, thanks to a patch originally by Pierre Mai.
3906 ** The system no longer bogusly warns about "abbreviated type
3908 ** The compiler gets less confused by inlining and RETURN-FROM,
3909 thanks to some patches originally by Tim Moore.
3910 ** The system no longer hangs when dumping circular lists to fasl
3911 files, thanks to a patch originally from Douglas Crosher.
3912 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3913 undefined function error.
3914 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3915 in structure slot initforms) for now, documented workaround instead:-|
3916 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3917 * fixed bug 5: added stubs for various Gray stream functions called
3918 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3919 installed, at least appropriate type errors are generated
3920 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3921 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3922 consistently in DEFMETHOD forms.
3923 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3924 it doesn't seem to affect SBCL after all
3925 * The C runtime system now builds with better optimization and many
3926 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3928 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3930 * The system is now under CVS at SourceForge (instead of the
3931 CVS repository on my home machine).
3932 * The new signal handling code has been tweaked to treat register
3933 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3934 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3935 downstream, e.g. in debug-int.lisp, has implicit dependencies
3936 on the unsignedness of integer representation of machine words,
3937 and that caused the system to bomb out with infinite regress
3938 when trying to recover from type errors involving signed values,
3939 e.g. (BUTLAST '(1 2 3) -1).)
3940 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3941 (This was one of the bugs Peter Van Eynde reported back in July.)
3942 * The system now uses code inspired by Colin Walters' O(N)
3943 implementation of MAP (from the cmucl-imp@cons.org mailing
3944 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3945 inline the MAP operation, and there is more than one
3946 sequence argument to the MAP call (so that it can't just
3947 do ETYPECASE once and for all based on the type of the
3948 single sequence argument). (The old non-inline implementation
3949 of the general M-argument sequence-of-length-N case required
3950 O(M*N*N) time when any of the sequence arguments were LISTs.)
3951 * The QUIT :UNIX-CODE keyword argument has been renamed to
3952 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3953 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3955 * Raymond Wiker's patches to port dynamic loading from Linux to
3956 FreeBSD have been added.
3957 * The BUGS file is now more nearly up to date, thanks in large part
3958 to Martin Atzmueller's review of it.
3959 * The debugger now flushes standard output streams before it begins
3960 its output ("debugger invoked" and so forth).
3961 * The core version number and fasl file version number have both
3962 been incremented, because of incompatible changes in the layout
3964 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3965 used to not be called for a saved Lisp image.)
3966 * Martin Atzmueller's version of a patch to fix a compiler crash,
3967 as posted on sbcl-devel 13 September 2000, has been installed.
3968 * Instead of installing Martin Atzmueller's patch for the
3969 compiler transform for SUBSEQ, I deleted the compiler transform,
3970 and transforms for some similar consing operations.
3971 * A bug in signal handling which kept TRACE from working on OpenBSD
3973 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3974 away by constant folding
3975 * The system now defines its address space constants in one place
3976 (in the Lisp sources), and propagates them automatically elsewhere
3977 (through GENESIS and the sbcl.h file). Therefore, patching the
3978 address map is less unnecessarily tedious and error-prone. The
3979 Lisp names of address space constants have also been systematized.
3980 * CVS tags like dollar-Header-dollar have been removed from
3981 the sources, because they have never saved me trouble and
3982 they've been source of trouble working with patches and other
3983 diff-related operations.
3984 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3985 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3987 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3989 * The system has been ported to OpenBSD.
3990 * The system now compiles with a simple "sh make.sh" on the systems
3991 that it's supported on. I.e., now you no longer need to tweak
3992 text in the target-features.lisp-expr and symlinks in src/runtime/
3993 by hand, the make.sh takes care of it for you.
3994 * The system is no longer so grossly inefficient when compiling code
3995 involving vectors implemented as general (not simple) vectors (VECTOR T),
3996 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3997 which dares to use the various sequence functions on non-simple
3998 vectors, takes less of a performance hit.
3999 * There is now a primitive type predicate VECTOR-T-P
4000 to test for the (VECTOR T) type, so that e.g.
4001 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4002 can now be compiled with some semblance of efficiency. (The old code
4003 turned the type declaration into a full call to %TYPEP at runtime!)
4004 * AREF on (VECTOR T) is still not fast, since it's still compiled
4005 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4006 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4007 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4008 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4009 without making full calls to SUBTYPEP at runtime.
4010 (Some analogous efficiency issues for non-simple vectors specialized to
4011 element types other than T, or for non-simple multidimensional arrays,
4012 have not been addressed. They could almost certainly be handled the
4013 same way if anyone is motivated to do so.)
4014 * The changes in array handling break binary compatibility, so
4015 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4016 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4017 instead of (VALUES T T).
4018 * By following the instructions that Dan Barlow posted to sbcl-devel
4019 on 2 July 2000, I was able to enable primitive dynamic object
4020 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4021 functionality is not implemented (since it calls ld to resolve
4022 library references automatically, requiring RUN-PROGRAM for its
4023 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4024 try to resolve library references) is now supported.
4025 * The system now flushes the standard output streams when it terminates,
4026 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4027 them at several other probably-convenient times, e.g. in each pass of
4028 the toplevel read-eval-print loop, and after evaluating a form given
4029 as an "--eval" command-line option. (These changes were motivated by a
4030 discussion of stream flushing issues on cmucl-imp in August 2000.)
4031 * The source transform for TYPEP of array types no longer assumes
4032 that an array whose element type is a not-yet-defined type
4033 is implemented as an array of T, but instead punts, so that the
4034 type will be interpreted at runtime.
4035 * There is now some support for cross-compiling in make.sh: each of
4036 the phases of make.sh has its own script. (This should be transparent
4037 to people doing ordinary, non-cross-compile builds.)
4038 * Since my laptop doesn't have hundreds of megabytes of memory like
4039 my desktop machine, I became more motivated to do some items on
4040 my to-do list in order to reduce the size of the system a little:
4041 ** Arrange for various needed-only-at-cold-init things to be
4042 uninterned after cold init. To support this, those things have
4043 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4044 symbols with such names are now uninterned after cold init).
4045 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4046 fasl files for cold load.
4047 ** Remove the old compiler structure pooling code (which used to
4048 be conditional on the target feature :SB-ALLOC) completely.
4049 ** Redo the representation of some data in cold init to be more compact.
4050 (I also looked into supporting byte compiled code at bootstrap time,
4051 which would probably reduce the size of the system a lot, but that
4052 looked too complicated, so I punted for now.)
4053 * The maximum signal nesting depth in the src/runtime/ support code has
4054 been reduced from 4096 to 256. (I don't know any reason for the very
4055 large old value. If the new smaller value turns out to break something,
4056 I'll probably just bump it back up.)
4057 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4059 * Many, many bugs reported by Peter Van Eynde have been added to
4060 the BUGS list; some have even been fixed.
4061 * While enabling dynamic object file loading, I tried to make the
4062 code easier to understand, renaming various functions and variables
4063 with less ambiguous names, and changing some function calling
4064 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4065 * While trying to figure out how to do the OpenBSD port, I tried to
4066 clean up some of the code in src/runtime/. In particular, I dropped
4067 support for non-POSIX signal handling, added various comments,
4068 tweaked the code to reduce the number of compilation warnings, and
4069 renamed some files to increase consistency.
4070 * To support the new automatic configuration functionality in make.sh,
4071 the source file target-features.lisp-expr has been replaced with the
4072 source file base-target-features.lisp-expr and the machine-generated
4073 file local-target-features.lisp-expr.
4074 * fixed a stupid quoting error in make.sh so that using CMU CL
4075 "lisp -batch" as cross-compilation host works again
4077 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4079 * DESCRIBE no longer tries to call itself recursively to describe
4080 bound/fbound values, so that it no longer fails on symbols which are
4081 bound to themselves (like keywords, T, and NIL).
4082 * DESCRIBE now works on generic functions.
4083 * The printer now prints less-screwed-up representations of closures
4084 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4085 * A private symbol is used instead of the :EMPTY keyword previously
4086 used to mark empty slots in hash tables. Thus
4087 (DEFVAR *HT* (MAKE-HASH-TABLE))
4088 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4089 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4090 now does what ANSI says that it should. (You can still get
4091 similar noncompliant behavior if bang on the hash table
4092 implementation with all the symbols you get back from
4093 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4094 This breaks binary compatibility, since tests for equality to
4095 :EMPTY are wired into things like the macroexpansion of
4096 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4098 * There's now a minimal placeholder implementation for CL:STEP,
4099 as required by ANSI.
4100 * An obscure bug in the interaction of the normal compiler, the byte
4101 compiler, inlining, and structure predicates has been patched
4102 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4103 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4105 * Missing ordinary arguments in a macro call are now detected even
4106 when the macro lambda list contains &KEY or &REST.
4107 * The debugger no longer complains about encountering the top of the
4108 stack when you type "FRAME 0" to explicitly instruct it to go to
4109 the top of the stack. And it now prints the frame you request even
4110 if it's the current frame (instead of saying "You are here.").
4111 * As specified by ANSI, the system now always prints keywords
4112 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4114 * The default initial SIZE of HASH-TABLEs is now smaller.
4115 * Type information from CLOS class dispatch is now propagated
4116 into DEFMETHOD bodies, so that e.g.
4117 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4119 is now basically equivalent to
4120 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4121 (DECLARE (TYPE SINGLE-FLOAT X))
4123 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4124 operation, without having to do run-time type dispatch.
4125 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4126 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4127 * Since I don't seem to be making big file reorganizations very often
4128 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4129 is overflowing my ability to conveniently back them up), I've finally
4130 checked the system into CVS. (The CVS repository is on my home system,
4131 not at SourceForge -- putting it on SourceForge might come later.)
4132 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4133 high-level GC-NOTIFY-FOO functions send their output. (There's
4134 still very little control of where low-level verbose GC functions
4135 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4136 less than it used to -- the GC-NOTIFY-FOO functions are now under
4137 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4138 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4139 in only one place in the source code, and propagates it automatically
4140 everywhere that it's needed. Thus e.g. when I bump the version from
4141 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4142 * The C source files now include boilerplate legalese and documentation
4143 at the head of each file (just as the Lisp source files already did).
4144 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4145 to his page will be replaced with a link to his new CLiki service.
4147 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4149 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4150 * The build process now looks for GNU make under the default name "gmake",
4151 instead of "make" as it used to. If GNU make is not available as "gmake"
4152 on your system, you can change this default behavior by setting the
4153 GNUMAKE environment variable.
4154 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4155 can build without error under CMU CL.
4157 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4159 * There is now a partial SBCL user manual (with some new text and some
4160 text cribbed from the CMU CL manual).
4161 * The beginnings of a profiler have been added (starting with the
4162 CMU CL profiler and simplifying and cleaning up). Eventually the
4163 main interface should be through the TRACE macro, but for now,
4164 it's still accessed through vaguely CMU-CL-style functions and macros
4165 exported from the package SB-PROFILE.
4166 * Some problems left over from porting CMU CL to the new
4167 cross-compilation bootstrap process have been cleaned up:
4168 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4169 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4170 part when getting rid of the compiler *BACKEND* stuff.)
4171 ** The value of *NULL-TYPE* was screwed up, because it was
4172 being initialized before the type system knew the final
4173 definition of the 'NULL type. This screwed up several key
4174 optimizations in the compiler, causing inefficiency in all sorts
4175 of places. (I found it because I wanted to understand why
4176 GET-INTERNAL-RUN-TIME was consing.)
4177 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4178 PROCLAIM FTYPE information. Unfortunately this broke binary
4179 compatibility again, since now the forms output by DEFGENERIC
4180 to refer to functions which didn't exist in 0.6.3.
4181 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4182 can use the new (as of 0.6.3) transform for SOME into MAP into
4184 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4185 values to (INTEGER 0 1000000), so that the time code will no longer
4186 occasionally get blown up by Linux returning 1000000 microseconds
4187 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4188 its output conform to the ANSI spec. (Alas, this makes its output
4189 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4190 * A full call to MAP NIL with a single sequence argument no longer conses.
4191 * fixes to problems pointed out by Martin Atzmueller:
4192 * The manual page no longer talks about multiprocessing as though
4193 it were currently supported.
4194 * The ILISP support patches have been removed from the distribution,
4195 because as of version 5.10.1, ILISP now supports SBCL without us
4196 having to maintain patches.
4197 * added a modified version of Raymond Toy's recent CMU CL patch for
4198 EQUALP comparison of HASH-TABLE
4200 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4202 * The system still can't cross-compile itself with
4203 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4204 entails), but at least it can compile more of itself that way
4205 than it used to be able to, and various buglets which were uncovered
4206 by trying to cross-compile itself that way have now been fixed.
4207 * This release breaks binary compatibility again. This time
4208 at least I've incremented the FASL file format version to 2, so that the
4209 problem can be detected reliably instead of just causing weird errors.
4210 * various new style warnings:
4211 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4212 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4213 ** using the lexical binding of a variable named in the *FOO* style
4214 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4215 as specified by ANSI.
4216 * *RANDOM-STATE* is no longer automatically initialized from
4217 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4218 default behavior of the system is to repeat its behavior every time
4219 it's run. If you'd like to change this behavior, you can always
4220 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4221 old convention there was no comparably easy way to get the system to
4222 repeat its behavior every time it was run.
4223 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4224 FTYPE declarations has been removed, in favor of their ANSI
4225 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4226 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4227 the types of their sequence arguments can be determined at compile time.
4228 This is done through a new open code expansion for MAP which eliminates
4229 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4230 argument types can be determined at compile time.
4231 * The optimizer now transforms COERCE into an identity operation when it
4232 can prove that the coerced object is already of the correct type. (This
4233 can be a win for machine generated code, including the output of other
4234 optimization transforms, such as the MAP transform above.)
4235 * Credit information has been moved from source file headers into CREDITS.
4236 * Source file headers have been made more standard.
4237 * The CASE macro now compiles without complaining even when it has
4240 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4242 * (Note that the way that the PCL macroexpansions were rewritten
4243 to accommodate the change in DEFGENERIC below breaks binary
4244 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4245 not run under sbcl-0.6.2. Once we get out of alpha releases,
4246 i.e. hit release 1.0.0, we'll probably try to maintain binary
4247 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4248 and sbcl-1.4.4. Until then, however, it might be fairly common
4249 for maintenance releases to break binary compatibility.)
4250 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4252 * The reporting of unhandled signals has been changed to print some
4253 explanatory text as well as the report form. (Previously only
4254 the report form was printed.)
4255 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4256 it defines, so that the compiler no longer issues undefined function
4257 warnings for compiled-but-not-yet-loaded generic functions.
4258 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4259 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4260 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4261 section of the ANSI spec are supported.
4262 * The "" nickname for the "KEYWORD" package has been removed.
4263 The reader still handles symbol tokens which begin with a package marker
4264 as keywords, but it doesn't expose its mechanism for doing so in the
4265 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4266 * The system now issues STYLE-WARNINGs for contradictory TYPE
4267 proclamations. (Warnings for contradictory FTYPE proclamations would
4268 be nice too, but those can't be done usefully unless the type system
4269 is made smarter about FUNCTION types.)
4270 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4271 systematized, so that "*target-*.lisp is supposed to exist only on the
4272 target and imply that there's a related file which exists on the
4273 host, and *host-*.lisp is supposed to exist only on the host and imply
4274 that there's a related file which exists on the target. This involves a
4275 lot of renaming. Hopefully the acute confusion caused by the renaming
4276 will be justified by the reduction in chronic confusion..
4277 ** runtime-type.lisp -> early-target-type.lisp
4278 ** target-type.lisp -> late-target-type.lisp
4279 ** early-host-format.lisp -> early-format.lisp
4280 ** late-host-format.lisp -> late-format.lisp
4281 ** host-error.lisp -> misc-error.lisp
4282 ** early-error.lisp -> early-target-error.lisp
4283 ** late-error.lisp -> late-target-error.lisp
4284 ** host-defboot.lisp -> early-defboot.lisp
4285 ** code/misc.lisp -> code/target-misc.lisp
4286 ** code/host-misc.lisp -> code/misc.lisp
4287 ** code/numbers.lisp -> code/target-numbers.lisp
4288 ** code/early-numbers.lisp -> numbers.lisp
4289 ** early-host-type.lisp -> early-type.lisp
4290 ** late-host-type.lisp -> late-type.lisp
4291 ** host-typep.lisp -> typep.lisp
4292 ** load.lisp -> target-load.lisp
4293 ** host-load.lisp -> load.lisp
4294 ** host-disassem.lisp -> disassem.lisp
4295 ** host-insts.lisp -> insts.lisp
4296 ** byte-comp.lisp -> target-byte-comp.lisp
4297 ** host-byte-comp.lisp -> byte-comp.lisp
4298 ** host-signal.lisp -> signal.lisp
4299 ** host-defstruct.lisp -> defstruct.lisp
4300 ** late-target-type.lisp -> deftypes-for-target.lisp
4301 Furthermore, several other previously target-only files foo.lisp (e.g.
4302 hash-table.lisp and random.lisp) have been split into a target-and-host
4303 foo.lisp file and a target-only target-foo.lisp file, with their key type
4304 definitions in the target-and-host part, so that the cross-compiler will
4305 know more about target types.
4306 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4307 gone away. In their place are various *BACKEND-FOO* variables
4308 corresponding to the slots of the old structure.
4309 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4310 into the target SBCL has been fixed.
4311 * The chill.lisp system for loading cold code into a running SBCL
4313 * Support for the CMU CL "scavenger hook" extension has been removed.
4314 (It was undocumented and unused in the CMU CL sources that SBCL was
4315 derived from, and stale in sbcl-0.6.1.)
4316 * Various errors in the cross-compiler type system were detected
4317 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4318 (enabling various consistency checks). Many of them were fixed,
4319 but some hard problems remain, so the compiler is back to
4320 running without *TYPE-SYSTEM-INITIALIZED* for now.
4321 * As part of the cross-compiler type system cleanup, I implemented
4322 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4323 * I have started adding UNCROSS calls throughout the type system
4324 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4325 on cross-compiler input files will be able to go away, and various
4327 * CONSTANTP now returns true for quoted forms (as explicitly required
4330 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4332 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4333 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4334 (MAKE-LIST -1). In the longer term, it ought to become true
4335 that declarations are assertions even at SAFETY 1. For now, it's not
4336 quite true even at SAFETY 3, but it's at least more nearly true..
4337 (Note that this change seems to increases the size of the system by
4338 O(5%) and to decrease the speed of the compiler by 20% or more.)
4339 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4340 for the problem of printing dozens of lines of distracting information
4341 about low-level system machinery as part of the top stack frame
4342 on entry to the debugger when an undefined function was called.
4343 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4344 is not reset to COMMON-LISP-USER.
4345 * Compilation of stuff related to dyncount.lisp has been made conditional
4346 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4347 smaller. The various dyncount-related symbols have been moved into
4348 a new "SB-DYNCOUNT" package.
4349 * tty-inspect.lisp has been renamed to inspect.lisp.
4350 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4351 feature has gone away. (When we eventually port to other flavors of
4352 libc and/or Unix, we'll try to make the differences between flavors
4353 invisible at the user level.)
4354 * Various other *FEATURES* tags, and/or their associated conditionals,
4355 have been removed if obsolescent, or given better documentation, or
4356 sometimes given more-mnemonic names.
4358 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4360 * tidied up "make.sh" script
4361 * tidied up system directory structure
4362 * better "clean.sh" behavior
4363 * added doc/FOR-CMUCL-DEVELOPERS
4364 * many many small tweaks to output format, e.g. removing possibly-confusing
4365 trailing #\. character in DESCRIBE-INSTANCE
4366 * (EQUALP #\A 'A) no longer signals an error.
4367 * new hashing code, including EQUALP hashing
4368 * tidied up Lisp initialization and toplevel
4369 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4370 * command line argument processing
4371 * added POSIX-GETENV function to deal with Unix-ish environment variables
4372 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4373 terminating SBCL on EOF
4374 * non-verbose GC by default
4375 * There is no more "sbcl" shell script; the sbcl file is now the C
4376 runtime executable (just like CMU CL).
4377 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4379 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4380 symbol and package information as Lisp native symbol and package objects
4381 instead of strings naming symbols and strings naming packages. This way,
4382 whenever packages are renamed (as in warm init), debug information is
4383 transformed along with everything else.
4384 * tweaked the optimization policy declarations which control the building
4385 of SBCL itself. Now, among other things, the system no longer saves
4386 source location debugging information. (This helps two problems at once
4387 by reducing SBCL size and by keeping SBCL from trying to look for its
4388 sources -- which may not exist -- when reporting errors.)
4389 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4390 debugging and testing purposes
4391 * cleaned up printing, making the printer call PRINT-OBJECT for
4392 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4393 methods, giving nearly-ANSI behavior
4394 * converted almost all special variables to use *FOO* naming convention
4395 * deleted PARSE-TIME functionality, since it can be done portably
4396 * moved some files out of cold init into warm init
4397 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4399 * regularized formatting of source files
4400 * added an install.sh script
4401 * fixed ridiculous memory usage of cross-compiler by making
4402 compiler/alloc.lisp not try to do pooling unless it can hook
4403 itself into the GC of the cross-compilation host. Now the system
4404 builds nicely on my old laptop.
4405 * added :SB-ALLOC in target-features.lisp-expr
4406 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4407 was not implemented)
4408 * re-did condition handling and note reporting in the compiler. Notes
4409 are no longer handled by signalling conditions. Style warnings
4410 and warnings are handled more correctly and reported in such a way
4411 that it's easy to find one or the other in your output (so that you
4412 can e.g. figure out which of many problems caused COMPILE-FILE to
4414 * changed the severity of several compiler warnings from full WARNING
4415 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4416 compiler note reporting so that it doesn't use the condition system
4417 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4418 COMPILE-FILE command)
4419 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4420 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4421 no longer supported, since their implementation was deeply intertwingled
4422 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4423 * removed bogus "support" for compiler macros named (SETF FOO), and
4424 removed the compiler macro for SETF INFO (but only after making a fool
4425 of myself on the cmucl-imp mailing list by posting a bogus patch for
4426 DEFINE-COMPILER-MACRO..)
4427 * Compiled files containing forms which have side effects on the Lisp
4428 reader (such as DEFPACKAGE forms) are now handled more correctly.
4429 (Compiler queuing of top level lambdas has been suppressed by setting
4430 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4431 * deleted various currently-unused source files, e.g. gengc.lisp. They
4432 may be added back at some point e.g. when porting to other architectures,
4433 but until they are it's distracting to distribute them and to try to
4435 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4436 there were so many of them they're just distractions, and UNCROSS is
4437 known to be able to handle the current sources
4438 * moved PROFILE functionality into TRACE, so that it will be clear
4439 how the wrapping and unwrapping of functions when you profile them
4440 interacts with the wrapping and unwrapping of functions when you
4441 trace them. (Actually, the functionality isn't there yet, but at least
4442 the interface specification is there. Hopefully, the functionality will
4443 arrive with some future maintenance release.)
4444 * removed host-oops.lisp
4445 * changed signature of QUIT function to allow UNIX-CODE argument
4446 * fixed READ-SEQUENCE bug
4447 * tweaked verbose GC output so that it looks more like the progress
4448 output that ANSI specifies for functions like LOAD
4449 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4450 * added <http://sbcl.sourceforge.com> to the banner information printed by