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. (spotted by a slight modification to some of PFD's tests)
35 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
36 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
37 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
38 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
40 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
41 and generic function now print the object identity for anonymous
42 instances. (thanks to Attila Lendvai)
43 * enhancement: as an extension to MOP, SBCL now supports SETF of
44 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
45 (thanks to Attila Lendvai)
46 * enhancement: sb-cover has an alternate annotation mode which puts
47 the form annotations at the car of the form, instead of annotating
48 it completely. The benefit of this mode is that it shows better
49 which forms were instrumented by the compiler.
50 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
51 by printer control variables. (reported by Dan Corkill)
52 * bug fix: system leaked memory when delayed output was performed by
53 the OS in smaller chunks then expected. (thanks to David Smith)
54 * bug fix: system leaked memory when file streams were not closed
56 * bug fix: large objects written to slow streams that were modified
57 after the write could end up with the modified state written to
58 the underlying file descriptor.
59 * bug fix: multiple threads operating in parallel on the same stream
60 could cause buffer-overflows.
61 * bug fix: source location information is stored correctly
62 (broken since 1.0.6). This bug would generally show up as the
63 Slime debugger higlighting the wrong form when the sldb-show-source
65 * bug fix: Branch forms are again annotated as branches in the sb-cover
67 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
68 on x86-64. (reported by Gregory Vanuxem)
69 * bug fix: EQUALP could return wrong results for structures with raw
70 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
71 word). (reported by Vjacheslav Fyodorov)
72 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
75 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
76 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
77 atomic compare-and-swap operations on threaded platforms.
78 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
79 allows assining a global minimum value to optimization qualities
80 (overriding proclamations and declarations).
81 * enhancement: closed over variables can be stack-allocated on x86
83 * performance bug fix: GETHASH and (SETF GETHASH) are once again
85 * optimization: slot definition lookup is now O(1). This speeds up
86 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
87 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
88 * optimization: ASSOC and MEMBER can now be open-coded for all
89 combinations of keyword arguments when second argument is constant
90 and SPEED >= SPACE. In other cases a specialized version is
92 * bug fix: using obsoleted structure instances with TYPEP and
93 generic functions now signals a sensible error.
94 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
95 (reported by Kristoffer Kvello)
96 * bug fix: backtrace construction is now more careful when making
97 lisp-objects from pointers on the stack, to avoid creating bogus
98 objects that can be seen by the GC.
99 * bug fix: defaulting of values in contexts expecting more than 7
100 variables now works on x86-64. (reported by Christopher Laux)
101 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
103 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
104 as the property-list of a symbol.
105 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
106 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
107 situations requested, are once again file-compileable. (reported
110 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
111 * MOP improvement: support for user-defined subclasses of
112 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
113 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
114 the expansion of DEFMETHOD, is responsible for generating a form
115 which creates a list of specializers when evaluated. Additional
116 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
117 debugging and introspective support.
118 * minor incompatible change: the (unsupported) spinlock interface
119 has changed: free spinlock now has the value NIL, and a held spinlock
120 has the owning thread as its value.
121 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
122 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
124 * enhancement: name of a socket-stream is now "a socket" instead of
126 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
127 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
128 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
129 for generic functions. (Thanks to Tobias C. Rittweiler)
130 * optimization: bignum printing speed has been improved by 20-40%
131 (depending on the bignum size.)
132 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
134 * bug fix: the cache used by the CLOS to store precomputed effective
135 methods, slot offsets, and constant return values is now thread and
137 * bug fix: generic function dispatch function updating is now thread
138 and interrupt safe (in the sense that the known issues have been
140 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
141 * bug fix: interning EQL-specializers is now thread and interrupt safe.
142 * bug fix: asdf systems with dependencies to the SB-POSIX or
143 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
144 * bug fix: interrupt safety of applicable method computation has been
147 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
148 * new contrib: sb-cover, an experimental code coverage tool, is included
150 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
151 significantly faster.
152 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
153 produce more efficient code on x86-64 if the compiler can determine
154 that (+ INDEX <constant>) does not require a bounds check and FOO
155 has an element type at least 8 bits wide.
156 * enhancement: a new, experimental synchronous timeout facility is
157 provided. Refer to SB-SYS:WITH-DEADLINE for details.
158 * enhancement: when a symbol name conflict error arises, the
159 conflicting symbols are always printed with a package prefix.
160 (thanks to Kevin Reid)
161 * enhancement: stepping is now once again supported on the SPARC and
162 MIPS platforms. (It is also now more likely to work on CheneyGC
164 * enhancement: sb-sprof can now also track and report accurate call
166 * bug fixes: the treatment of non-standard subclasses of
167 SB-MOP:SPECIALIZER is more correct.
168 * incompatible change: PURIFY no longer copies the data from the
169 dynamic space into the static and read-only spaces on platforms
170 that use the generational garbage collector
171 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
173 * bug fix: GC race condition occasionally resulting in crashes with
174 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
176 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
178 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
179 a STYLE-WARNING to be signalled (regression from 1.0.4.)
180 * bug fix: an asynchronous interrupt could previously leave the
181 system running with GC inhibited.
182 * bug fix: a DECLARE form evaluated at top-level now causes an error
183 rather than silently (or verbosely) returning NIL.
184 * bug fix: trying to dissassemble functions compiled at high DEBUG could
185 result in a error being signalled due to source form lookup errors.
186 (reported by Peter Graves)
188 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
189 * incompatible change: removed writer methods for host-ent-name,
190 host-ent-addresses -- changing the values did not update the DNS
192 * minor incompatible change: changed experimental JOIN-THREAD interface
193 * documentation: the manual now lists reader and writer methods
194 in class slot documentation sections. (thanks to Richard M Kreuter)
195 * documentation: unwinding from asyncronous events has been
196 documented as unsafe.
197 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
198 in multithreaded application code.
199 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
200 platforms. (thanks to James Anderson for the optimization hint)
201 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
203 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
204 variants no longer cons.
205 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
206 their NOT- variants no longer cons.
207 * optimization: EQUAL hash tables no longer use SXHASH for objects
208 of all data types, but instead use an EQL hash for types for which
209 EQUAL is the same as EQL.
210 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
211 are significantly faster.
212 * optimization: new STRING-OUTPUT-STREAM implementation is much
213 faster for large outputs and conses 30% less on average.
214 * enhancement: XREF information is now collected for references made
215 to global variables using SYMBOL-VALUE and a constant argument.
216 * enhancement: SIGINT now causes a specific condition
217 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
218 * bug fix: FILE-POSITION used to signal an error in some cases where
219 ANSI requires it to return NIL.
220 * bug fix: ADJUST-ARRAY is now interrupt-safe.
221 * bug fix: adding and removing fd-handlers is now interrupt-safe.
222 * bug fix: inlined calls to C now ensure 16byte stack alignment on
224 * bug fix: bad type declaration in the CLOS implementation has
225 been fixed. (reported by James Anderson)
226 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
227 (thanks to Lutz Euler)
228 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
229 required by the CLHS. (reported by Samium Gromoff)
230 * bug fix: dead unbound variable references now signal an error.
231 (reported by Marco Monteiro)
232 * bug fix: / with an unused value was being deleted in safe code.
233 (thanks to Marco Monteiro and Kevin Reid)
234 * bug fix: number of characters that can be written onto a single
235 line in a file is unlimited.
236 * bug fix: some GC deadlocks caused by asynchronous interrupts have
237 been fixed by inhibiting interrupts for when GC is disbled.
238 * bug fix: some interrupt safety issues with GETHASH, (SETF
239 GETHASH), CLRHASH and REMHASH have been fixed.
240 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
241 specifier no longer causes infinite recursion.
242 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
243 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
244 * bug fix: an &environment argument with the correct variable information
245 is passed to macros that are expanded during byte compilation.
246 (reported by Samium Gromoff)
247 * bug fix: a logic error was causing the hash function for bignums to
248 have excessive amounts of collisions. (reported by Faré Rideau)
249 * bug fix: modifying the contents of an array could change the return
250 value of SXHASH on that array, which is only allowed for strings
251 and bit vectors. (bug introduced in 0.9.16)
252 * improvement: the x86-64/darwin port now passes all tests (except
253 for the debugger tests) but should still be considered
254 experimental until this is fixed.
255 * improvement: a style-warning is signaled for CASE (etc) clauses with
256 duplicate keys. (thanks to Kevin Reid)
257 * improvement: macos/x86 and macos/x86-64 now use mach exception
258 handlers for dealing with illegal instructions (for trapping and
259 error handling) and memory protection violations (for GC).
261 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
262 * new platform: experimental support for x86-64/darwin (MacOS).
263 * incompatible change: the thread-safe (on most platforms) getaddrinfo
264 and getnameinfo sockets functions are used instead of gethostbyaddr
265 and gethostbyname, on platforms where the newer functions are available.
266 As a result, the ALIASES field of HOST-ENT will always be NIL on these
268 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
269 don't print the startup banner, but behave as if --noinform was passed
270 as a command line argument. (thanks to Kevin Reid)
271 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
272 * optimization: code using alien values with undeclared types is much faster.
273 * optimization: the compiler is now able to open code SEARCH in more cases.
274 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
275 * bug fix: using standardized COMMON-LISP special variables as loop
276 variables no longer signals bogus package lock violations. (reported
278 * bug fix: declaring local loop variables to be of a range-limited type
279 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
280 (reported by Andras Simon)
281 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
282 bugs remain on x86-64.)
283 * bug fix: the #= and ## reader macros now interact reasonably with
284 funcallable instances.
285 * bug fix: type-checks for function arguments were compiled using the
286 compiler policy settings of the wrong lexical environment (in 1.0.2
288 * bug fix: SHADOW accepts characters as string designators, as required
289 by the spec (thanks to Eric Marsden)
290 * bug fix: fixed GC safety issues when foreign functions are called with
291 non-base strings as arguments
292 * bug fix: more consistent error output for fatal-compiler-warnings, like
294 * bug fix: functions with proper names were showing up as NIL in
295 backtraces (reported by Edi Weitz, regression in 1.0.2)
297 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
298 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
299 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
300 produce more efficient code on the x86 if the compiler can determine
301 that (+ INDEX <constant>) does not require a bounds check and FOO
302 has an element type at least 8 bits wide.
303 * bug fix: references to undefined variables are handled the same way
304 in toplevel forms as in the normal compiler
305 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
307 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
308 (thanks to Jon Buller)
309 * improvement: faster compilation times for complex functions
310 * improvement: added readlink support to SB-POSIX (thanks to Richard
313 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
314 * improvement: experimental support for mach exception handling on
315 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
317 * improvement: support for GBK external format.
318 (thanks to Chun Tian (binghe))
319 * improvement: the debugger now displays variables that have been closed
320 over, in code compiled with (DEBUG 2) or higher
321 * improvement: support for executable cores on NetBSD (thanks to
323 * new feature: added a RESTART-FRAME debugger command
324 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
325 be used to provide an implementation for FILE-POSITION on Gray streams
326 (thanks to Eric Marsden)
327 * new feature: add syslog, openlog and closelog support to SB-POSIX
328 (thanks to Richard Kreuter)
329 * optimization: the function call overhead in code compiled with
330 a high DEBUG optimization setting is significantly
331 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
332 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
333 (thanks to Tony Martinez)
334 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
335 evaluated once (reported by Kevin Reid)
336 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
337 works again (thanks to Kilian Sprotte)
338 * bug fix: an error is signaled for tagbodies with duplicate tags
339 (thanks to Stephen Wilson)
340 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
341 * bug fix: Win32 port can now handle foreign code unwinding Lisp
342 stack frames from alien callbacks.
343 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
344 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
345 65k pages (thanks to David Woodhouse)
346 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
348 changes in sbcl-1.0.1 relative to sbcl-1.0:
349 * new platform: FreeBSD/x86-64, including support for threading.
350 * new feature: the compiler stores cross-referencing information
351 abount function calls (who-calls), macroexpansion (who-macroexpands)
352 and special variables (who-binds, who-sets, who-references) for code
353 compiled with (< SPACE 3). This information is available through the
354 sb-introspect contrib.
355 * new feature: users may subclass SEQUENCE, and have instances of
356 these classes interoperate with standard Common Lisp functions if
357 a number of methods are defined. (This feature is experimental
358 and the interface subject to change based on feedback from SBCL
359 users and the general community)
360 * improvement: sb-sprof traces call stacks to an arbitrary depth on
361 x86/x86-64, rather than the previous fixed depth of 8
362 * improvement: another pthread back-end of mutex "pthread-futex".
363 * bug fix: non-ascii command-line arguments are processed correctly
364 (thanks to Yaroslav Kavenchuk)
365 * bug fix: non-required arguments were not passed correctly when a method
366 defined using DEFMETHOD was called from a mop-generated method using
367 CALL-NEXT-METHOD (reported by Pascal Costanza)
368 * bug fix: recursion is now permitted in accessors through
369 SLOT-UNBOUND. (reported by Pascal Costanza)
370 * bug fix: an error was signaled at startup if the HOME environment
371 variable was defined, but had an empty value (reported by Peter Van Eynde)
372 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
373 signal a PROGRAM-ERROR, not a BUG.
374 * bug fix: SB-POSIX:READDIR works when built with large file support.
375 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
377 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
378 for better automatic indentation support. (thanks to Matt Pillsbury)
379 * optimization: loading generic functions no longer takes O(n^2) time,
380 proportional to the amount of methods in the generic function
381 (reported by Todd Sabin and Jeremy Brown)
382 * optimization: the FIND and POSITION family of sequence functions
383 are significantly faster on arrays whose element types have been
385 * improvements to the Windows port:
386 ** Intermittent heap corruption problems have been fixed. (thanks
387 to Alastair Bridgewater)
388 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
390 ** Lisp is able to unwind foreign exception frames from alien
391 callbacks. (thanks to Alastair Bridgewater)
393 changes in sbcl-1.0 relative to sbcl-0.9.18:
394 * improvement: experimental support for threading on FreeBSD/x86.
395 (thanks to NIIMI Satoshi)
396 * improvement: runtime option --dynamic-space-size can be used
397 to set the size of the dynamic space reserved on startup.
398 * improvement: floating point modes in effect are now saved in
399 core, and restored on startup.
400 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
401 startup, not time since first call to GET-INTERNAL-REAL-TIME.
402 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
403 threads are not running after *SAVE-HOOKS* have run.
404 * improvement: writes to CLOS instance slots are type-checked in code
405 compiled with (SAFETY 3)
406 * improvement: floating-point exception handling on FreeBSD (thanks to
408 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
409 (thanks to Zach Beane)
410 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
412 * improvement: added support for the Shift-JIS external format.
413 (contributed by NIIMI Satoshi)
414 * improvement: callbacks are supported on Linux/PPC. (thanks to
416 * bug fix: compiler bug triggered by a (non-standard) VALUES
417 declaration in a LET* was fixed. (reported by Kaersten Poeck)
418 * bug fix: file compiler no longer confuses validated and already
419 dumped structurres. (reported by Kaersten Poeck)
420 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
421 fill-pointer signals a type-error as required. (thanks to
423 * bug fix: disassemly of funcallable instances works.
424 * bug fix: single stepping on PPC.
425 * bug fix: fix thread-safety problems in the type system (generally
426 manifesting as nonsensical errors like "STRING is a bad type specifier
427 for sequences" or "The value 1 is not of type FIXNUM")
428 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
429 (reported by Josip Gracin)
430 * bug fix: an error is signaled for attempts to displace arrays with
431 incompatible element types (thanks to Mario Mommer)
432 * bug fix: more correct handling of wide characters in debug info
433 (bug reported by Attila Lendvai and fixed by Juho Snellman)
434 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
435 and don't cause extra consing
436 * optimization: MAP and MAP-INTO are significantly faster on vectors
437 whose elements types have been declared.
438 * Improvements to SB-SPROF:
439 ** Support for allocation profiling
440 ** Reduced profiling overhead, especially for long profiling runs
441 * Improvements to the Windows port:
442 ** floating point exceptions are now reported correctly.
443 ** stack exhaustion detection works partially.
444 ** more accurate GET-INTERNAL-REAL-TIME.
445 ** better RUN-PROGRAM behaviour in case of errors in the child
447 ** PROBE-FILE now simplifies pathnames correctly.
448 ** DIRECTORY now works correctly with :WILD-INFERIORS.
449 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
450 1st 1970 (although time timezone data will be unreliable).
451 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
453 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
454 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
455 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
457 * bug fix: two potential GC deadlocks affecting threaded builds.
458 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
459 returns T (reported by Anton Kazennikov)
460 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
461 with non-variable places
462 * bug fix: remove a race condition in the setting of
463 funcallable-instance functions, this should make threaded CLOS
464 code more stable against memory faults.
465 * bug fix: corruption of specials when unbinding is interrupted by an
466 asynchronous unwind (reported by Hannu Koivisto)
467 * improvement: the debugger will now also display local variables that
468 are only used once, for code compiled with a DEBUG optimization quality
471 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
472 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
473 * incompatible change: External-format support for FFI calls. The
474 SB-ALIEN:C-STRING no longer implies an ASCII
475 external-format. Instead the string is subject to external-format
476 conversion. Additionally return values of type SB-ALIEN:C-STRING
477 are by default of type (SIMPLE-ARRAY CHARACTER), not
478 SIMPLE-BASE-STRING. For an alien type that behaves like the old
479 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
480 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
481 * incompatible change: SB-EXT package no longer contains the
482 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
483 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
485 * incompatible change: the single-stepper is no longer available
486 on Alpha, Mips and Sparc platforms.
487 * minor incompatible change: the direct superclasses of
488 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
489 not (STANDARD-OBJECT FUNCTION). This makes the
490 class-precedence-lists of GENERIC-FUNCTION and
491 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
493 * minor incompatible change: the default stream external format on
494 non-unicode SBCL is no longer detected from the locale, but is always
496 * new feature: Add a version of evaluator that uses an interpreter instead
497 of the compiler. EVAL still uses the compiler by default, to switch it
498 to use the interpreter, set the value of the variable
499 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
500 * minor incompatible change: the single-stepper REPL has been merged
501 with the normal debugger (see the "Stepping" heading of the debugger help
502 for more details). The debugger command STEP will no longer switch
503 to the single-stepper REPL.
504 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
505 for a type now works.
506 * bug fix: loading of default sysinit file works. (thanks to Leonid
508 * bug fix: better detection of circularities in the file-compiler.
509 (reported by Marco Monteiro)
510 * bug fix: the CL pathname functions now work with files that have
511 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
512 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
513 systems with Unix98 pty semantics.
514 * bug fix: ASDF-INSTALL will now work with bsd tar.
515 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
517 * bug fix: timers expiring in dead threads no longer cause a
518 type-error (reported by Paul "Nonny Mouse").
519 * bug fix: thanks to more lightweight single-stepper instrumentation,
520 code compiled with (DEBUG 3) will compile and execute significantly faster,
521 and will have more accurate type-inferencing than before
522 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
523 whose bindings are modified
524 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
525 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
526 ** replace dummy implementations of CL:MACHINE-INSTANCE and
527 CL:SOFTWARE-VERSION with proper version
529 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
530 * feature: implemented the READER-METHOD-CLASS and
531 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
532 as specified by AMOP.
533 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
535 * optimization: faster LOGCOUNT implementation on x86 and x86-64
536 (thanks to Lutz Euler)
537 * optimization: hashing of general arrays and vectors has been
538 improved. (reported by Any Fingerhut)
539 * enhancement: SB-INTROSPECT is now able to find definitions of
540 profiled functions. (thanks to Troels Henriksen)
541 * enhancement: compiler-macro expansion applies now to FUNCALL forms
542 as well, allowing compiler-macros for SETF-functions to expand.
543 * enhancement: step-instrumentation no longer wraps known
544 single-value functions with multiple-value context, allowing
545 better type inference.
546 * fixed bug #337: use of MAKE-METHOD in method combination now works
547 even in the presence of user-defined method classes. (reported by
548 Bruno Haible and Pascal Costanza)
549 * fixed bug #339(c): if there are applicable methods not part of any
550 long-form method-combination group, call INVALID-METHOD-ERROR.
551 (reported by Bruno Haible)
552 * fixed bug #361: the :FUNCTION initarg in the protocol for
553 initialization of methods can now be used to override
554 internally-produced optimized functions. (reported by Bruno
556 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
557 system-provided lambda expression no longer cause warnings about
558 unbound #:|pv-table| symbols.
559 * bug fix: improved the handling of type declarations and the
560 detection of violations for keyword arguments with non-constant
562 * bug fix: potentially erronous calls to PATHNAME and
563 MERGE-PATHNAMES were being flushed in some cases.
564 (reported by Richard Kreuter)
565 * bug fix: compiled calls to TYPEP were mishandling obsolete
566 instances. (reported by James Bielman and Attila Lendvai)
567 * bug fix: format strings with ~> without matching ~< no longer
568 trigger an AVER, but signal an understandable error instead.
569 (reported by Antonio Martinez)
570 * bug fix: specifying an output-file without a pathname-type for
571 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
572 of input-file instead of "fasl". (reported by Robert Dodier)
573 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
574 (reported by James Y Knight).
575 * bug fix: compiler-macros-function did not consider the environment
576 argument for shadowing by local functions.
577 * bug fix: compiler-macros expansion was inhibited by local INLINE
579 * bug fix: inline expansions of known functions were subject to
580 step-instrumentation in high debug policies, leading to problems
582 * bug fix: compiler failed to differentiate between different CONS
584 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
585 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
586 * bug fix: #368: incorrect use of expressed vs. upgraded array
588 * bug fix: #306a: more precise unions of array types.
589 * thread-safety improvements:
590 ** CONDITION-WAIT could return early on Linux, if the thread was
591 interrupted and subsequently continued with SIGCONT.
592 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
594 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
595 * added support for the ucs-2 external format. (contributed by Ivan
597 * minor incompatible change: pretty printing of objects of type
598 (cons symbol) is, in the default pprint-dispatch-table, now
599 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
601 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
602 called later in a class's lifetime, possibly as late as when the
603 first instance of the class is created. Previously,
604 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
605 class became finalizeable.
606 * fixed bug: FILE-POSITION sometimes returned inconsistent results
607 for multibyte external-format streams. (thanks to "vbzoli")
608 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
609 with :ALLOCATION :CLASS inherited from superclasses of the
611 * fixed bug: anonymous classes can now be created using the :NAME
612 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
613 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
614 * fixed bug: core-files saved with :EXECUTABLE T can again be
615 executed when SBCL_HOME isn't set. (reported by James Knight)
616 * fixed bug: toplevel LOCALLY forms with declarations could
617 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
618 * fixed bug: printing from several different threads using different
619 values of *print-case* could cause invalid output, due to
620 some internal special variables of the printer not being bound
621 thread-locally (reported by Max Mikhanosha)
622 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
623 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
624 (reported by Pascal Costanza)
625 * fixed bug: FINALIZE-INHERITANCE is called from
626 REINITIALIZE-INSTANCE on classes when the class has previously
627 been finalized, as required by AMOP.
628 * minor code generation optimizations:
629 ** better register allocation in CLOS dispatching functions
630 ** overflow detection when coercing signed bytes to fixnums on x86-64
631 ** is now implemented with one IMUL instruction instead of three shifts
632 ** more efficient bit-vector access on x86 and x86-64
633 ** more efficient access to raw structure slots on x86 and x86-64
634 * fixed some bugs revealed by Paul Dietz' test suite:
635 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
638 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
639 * feature: thread support on Solaris/x86, and experimental thread support
641 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
643 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
644 default method for SLOT-UNBOUND.
645 * minor incompatible change: prevent the user from specializing the
646 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
647 somewhat counter to the intent of the protocol, I (CSR) think, and
648 additionally it just doesn't work in SBCL as currently
649 implemented, thanks to optimizations (that are always valid for
650 the other three SLOT-VALUEish functions, but not for the setter).
651 * minor incompatible change: the :SB-LDB feature is now enabled by
652 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
653 the low-level debugger.
654 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
655 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
656 down to an SB-SYS:FD-STREAM, allowing direct communication in
657 more cases. Temporary files and pipes are still used as a fallback
659 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
660 on the non-reentrant C functions localtime(3) and gmtime(3).
661 * bug fix: global optimization policy was not visible in LOCALLY and
663 * bug fix: class objects can be used as specializers in methods.
664 (reported by Pascal Costanza)
665 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
666 with a type error. (reported by blitz_ on #lisp)
667 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
668 failed with a type error. (reported by Pascal Bourguignon)
669 * bug fix: merging pathnames against defaults with :DIRECTORY
670 starting with '(:RELATIVE :BACK) should preserve the :BACK.
671 (reported by James Y Knight)
672 * bug fix: saving large (>2GB) cores on x86-64 now works
673 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
674 constant index and a (simple-array (signed-byte 32)) array
675 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
676 error (patch by Robert J. Macomber)
677 * feature: TIME now displays the amount of run-time spent in GC
678 * bug fix: The debugger now does a better job of respecting
679 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
680 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
681 workaround for bug 403.)
682 * bug fix: floating point exception handling now works on FreeBSD
683 (thanks to NIIMI Satoshi)
684 * fixed some bugs revealed by Paul Dietz' test suite:
685 ** MISC.641: LET-conversion were not supposed to work in late
687 * improvements to the Win32/x86 port:
688 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
689 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
691 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
692 * new feature: source path information is generated for macro-expansion
693 errors for use in IDE's like Slime (thanks to Helmut Eller)
694 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
695 * bug fix: compiler does not loop forever on an invalid type in
697 * improvement: compilation of most CLOS applications is significantly
699 * optimization: added a limited bytecode compiler for simple toplevel
700 forms, speeding up compilation and FASL loading
701 * bug fix: the statistical profiler now properly distinguishes anonymous
704 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
705 * minor incompatible change: in sbcl-0.9.11 (but not earlier
706 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
707 into subclasses of the finalized class. Now user calls to
708 FINALIZE-INHERITANCE finalize just the one class, and calls by the
709 system return before any subclasses are finalized.
710 * minor incompatible change: The reader no longer ignores errors
711 regarding non-existent packages in #+ and #- feature tests.
712 * new feature: command line options --no-sysinit, --no-userinit to
713 inhibit loading the corresponding init files
714 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
715 for sending data through UDP sockets (thanks to François-René Rideau)
716 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
717 error is signalled instead (thanks to François-René Rideau)
718 * minor incompatible change: Error signalling behaviour of lexical
719 operations violating package locks has changed slightly. Refer to
720 documentation on package locks for details.
721 * enhancement: EVAL can process IF-expressions without resorting to the
723 * bug fix: Some operations involving SIN, COS, or TAN weren't being
724 constant-folded properly due to missing out-of-line functions.
725 (reported by Mika Pihlajamäki)
726 * bug fix: LISTEN sometimes returned T even in cases where no data was
727 immediately available from the stream
728 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
729 were reversed. (reported by Levente Mészáros)
730 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
731 when given a vector displaced to an adjustable array. (reported
733 * bug fix: derivation of float boundaries from numbers outside the
734 appropriate float range (reported by John Wiseman)
735 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
737 * bug fix: precomputing cache entries for generic functions with
738 some subclasses of specializers as yet invalid does not attempt to
739 fill a cache line with a negative offset. (reported by Levente
741 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
742 allow condition class objects as arguments to DOCUMENTATION and
743 (SETF DOCUMENTATION); only find and set documentation for
744 structure names for the STRUCTURE doc-type. (suggested by Gary
746 * improvements to the Win32/x86 port:
747 ** (user-homedir-pathname) and default initialization file
748 locations now know about the user's "Documents and Settings"
749 directory (thanks to Yaroslav Kavenchuk)
750 ** run-program is implemented (thanks to Mike Thomas)
751 ** sockets support (thanks to Timothy Ritchey)
752 ** better backtrace support (thanks to Alastair Bridgewater)
753 ** sb-grovel supported
754 ** asdf-install and sb-posix work somewhat
755 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
756 * improvements to the Solaris/x86 port:
757 ** works on Solaris 11/Solaris Express
758 ** floating-point exception handling support
759 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
760 * fixed some bugs revealed by Paul Dietz' test suite:
761 ** REMOVE-METHOD returns its generic function argument even when
762 the method is not one of the generic functions' methods.
763 ** objects declared as MEMBER types can be admissible arguments to
765 ** printing characters should simply be printed by the FORMAT ~:C
767 ** compiler failure when compiling functions with hairy constant
768 defaults for optional parameters.
769 ** compiler produces wrong code when MAYBE-INLINE-expanding a
770 function, which is already optimized.
772 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
773 * new platform: experimental support for SBCL x86/Darwin, including
774 MacOS X 10.4.5 on Intel.
775 * new feature: Unicode character names are now known to the system
776 (through CHAR-NAME and NAME-CHAR).
777 * new feature: the filesystem location of SBCL's core file is
778 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
779 * minor incompatible change: the contrib modules SB-POSIX and
780 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
781 this change is to make it easier to distribute
782 location-independent binaries.
783 * bug fix: as implied by AMOP, standardized classes no longer have
784 slots named by external symbols of public packages. (reported by
786 * optimization: calling functions via constant symbols -- as in
787 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
788 via the function object as in (FUNCALL #'FOO).
789 * optimization: on x86 and x86-64, the calling convention for the
790 case of dealing with an unknown number of values has been altered
791 to be friendlier to the prediction heuristics implemented,
792 particularly with reference to CALL and RET pairing. (thanks to
793 Alastair Bridgewater)
794 * optimization: on x86, the code for access to thread-local symbol
795 values and binding/unbinding of thread-local symbols is smaller.
796 (thanks to Alastair Bridgewater)
797 * enhancement: CONSTANTP is now able to determine constantness of
798 more complex forms, including calls to constant-foldable standardized
799 functions and some special forms beyond QUOTE.
800 * fixed bug: occasional GC crashes on Solaris/x86
801 * optimization: x86-64 supports stack allocation of results of simple
802 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
803 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
804 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
805 (thanks to James Knight)
806 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
807 by the standard (thanks to Stephen Compall)
809 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
810 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
811 be used for bundling the runtime and the core file into one
812 executable binary. This feature is not currently supported on all SBCL
813 platforms. (thanks to James Bielman and NIIMI Satoshi)
814 * new feature: a generational or ephemeral garbage collector is now
815 the default on the PowerPC platform (both Linux and Darwin). The
816 old Cheney (stop and copy) collector is a build-time option.
817 (thanks to Cyrus Harmon, after Raymond Toy)
818 * minor incompatible change: the method by which SBCL finds its
819 contributed modules has changed; it no longer relies on symbolic
820 links from an $SBCL_HOME/systems directory, but searches directly
821 in the subdirectories of $SBCL_HOME.
822 * enhancement: the dynamic heap size on the Linux/PPC platform is
823 markedly larger, even using the older Cheney garbage collector.
824 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
825 certain complicated string types. (reported by Gary King)
826 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
827 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
828 * fixed bug: DOCUMENTATION on structures no longer signals an
829 error if no structure type of the right name exists. (reported by
831 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
832 applicable non-standard primary or :AROUND method on
833 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
834 no longer cause unbound variable errors. (reported by Kevin Reid)
835 * fixed bug: metacircle resolution in cases where methods have slots
836 added before the slots from STANDARD-METHOD. (reported by Jean
838 * fixed bug: the Power PC assembly code for calling into lisp
839 sometimes computed its return address wrongly (depending on the
840 15th bit of the address, fixed at link time). This used to
841 manifest itself in a segmentation violation while building PCL.
842 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
843 many others over the years)
844 * fixed bug: ROOM no longer reports silly numbers for stack usage if
845 the stack pointer (treated as a signed integer) is negative.
846 (thanks to Peter van Eynde)
848 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
849 * new platform: experimental support for the Windows operating
850 system has been added. (thanks to Alastair Bridgewater)
851 * fixed several bugs in and robustified the PPC FFI (including
852 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
853 * bug fix: allow non-simple string symbol names (reported by Paul
855 * bug fix: interrupt handling on NetBSD (thanks to Richard M
857 * bug fix: saving a core corrupted callbacks on x86/x86-64
858 * bug fix: closed a loophole in metacircularity detection and
859 grounding in the PCL implementation of CLOS.
860 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
861 (reported by tomppa on #lisp)
862 * bug fix: clear the direction flag on Lisp -> C transitions, as
863 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
864 (reported by Andrej Grozin and Hendrik Maryns)
865 * optimization: major improvements to GC efficiency on GENCGC platforms
866 * optimization: faster implementation of EQUAL
867 * optimization: emit more efficient opcodes for some common
868 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
869 * optimization: several other minor code-generation improvements on x86-64
870 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
872 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
873 * minor incompatible change: (SETF CLASS-NAME) and (SETF
874 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
875 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
876 by ANSI as a generic function, and for consistency (SETF
877 GENERIC-FUNCTION-NAME) is treated likewise.
878 * fixed bug #233.b: make constraint propagation notice when a variable
879 value is changed after it is referenced but before it is used
880 * fixed bug #296: no more arbitrary behaviour from filenames with
881 odd characters as --load arguments. (reported by Adam Warner)
882 * fixed bug #390: :CHARACTER-SET pathname components now work as
883 expected. (reported by Tim Daly Jr)
884 * fixed bug #391: complicated :TYPE intersections in slot
885 definitions no longer cause an error in PCL internals.
886 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
887 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
888 returning the number of octets which would be written to the
889 file-stream. (thanks to Robert J. Macomber)
890 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
891 arguments to the internal stream functions. (thanks to David
893 * bug fix: stack exhaustion is now handled on the x86/SunOS
894 platform. (thanks to Daisuke Homma)
895 * bug fix: more accurate ROOM results on GENCGC platforms
896 * optimization: improved type inference for arithmetic-for
897 index variables in LOOP
898 * optimization: faster floating-point SQRT on x86-64
899 * fixed some bugs revealed by Paul Dietz' test suite:
900 ** DOCUMENTATION returns NIL instead of "" for method combinations
901 that don't have a docstring
903 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
904 * minor incompatible change: (SETF CLASS-NAME) and (SETF
905 GENERIC-FUNCTION-NAME) are no longer generic functions, and
906 therefore are not customizeable by user code (as seems to be at
907 least permitted and maybe required by AMOP). As a consolation,
908 however, the SBCL implementation of these functions now calls
909 REINITIALIZE-INSTANCE as specified by AMOP.
910 * bug fix: slot-definition documentation is propagated as per ANSI
911 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
912 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
913 argument, as do eslotd initialization methods. (from Pascal
914 Costanza's "Closer" project)
915 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
916 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
918 * bug fix: it is now possible to have more than one subclass of
919 STANDARD-GENERIC-FUNCTION without causing stack overflow.
920 (reported by Bruno Haible, Pascal Costanza and others)
921 * bug fix: the dependent update protocol now works for generic
922 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
924 * bug fix: condition-class instances corresponding to
925 DEFINE-CONDITION forms are now created eagerly. (reported by
926 Kalle Olavi Niemitalo on comp.lang.lisp)
927 * bug fix: floating point printing is more accurate in some
928 circumstances. (thanks to Simon Alexander)
929 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
930 merged with *DEFAULT-PATHNAME-DEFAULTS*.
931 * bug fix: callbacks on OS X now preserve stack-alignment.
932 (thanks to Cyrus Harmon)
933 * enhancement: the x86-64 disassembler is much better at
934 disassembling SSE instructions. (thanks to Lutz Euler)
935 * enhancement: CHECK-TYPE on a variable now allows type-inference
936 (thanks to Vincent Arkesteijn)
937 * optimization: improved performance of EUC-JP external format.
938 (thanks to NIIMI Satoshi)
939 * optimization: performance improvements to IO on file streams of
940 :ELEMENT-TYPE CHARACTER
941 * optimization: much faster memory allocation on x86-64
942 * optimization: faster garbage collections (latency and throughput)
944 * optimization: faster bignum subtraction and division on x86 and x86-64
945 (thanks to Lutz Euler)
946 * optimization: more accurate type-inference for loops that use a
947 floating point index variable or a negative step.
949 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
950 * bug fix: add a workaround to SBCL looping infinitely at startup on
951 Linux kernels with apparently buggy implementations of personality().
952 (thanks to Svein Ove Aas)
953 * bug fix: Unicode symbols are correctly printed in LDB backtraces
954 (thanks to David Lichteblau)
955 * bug fix: local bindings shadow global symbol macros for
956 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
957 * optimization: non-open coded uses of numeric comparison operators
958 (e.g. >) no longer cons when called with more than one parameter
959 on platforms supporting dynamic-extent allocation.
960 * enhancement: saving cores with foreign code loaded is now supported
961 on MIPS/Linux in addition to the previously supported platforms.
962 * bug fix: FIND-RESTART now tests for activity, not applicability when given
963 a restart object as identifier. (reported by Helmut Eller for CMUCL)
964 * bug fix: division by zero in sb-sprof when no samples were collected
965 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
966 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
967 (thanks to Alexander Kjeldaas)
968 * new feature: ignoring the return values of destructive functions
969 like NREVERSE signals a compile-time style-warning.
970 (thanks to Kevin Reid)
971 * new feature: the alignment of alien structure fields can be
972 explicitly specified. (thanks to Cyrus Harmon)
973 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
974 and dump core on SIGQUIT
976 ** incompatible change: threads do not inherit values of specials
977 from their parents (see manual)
978 ** bug fix: threads stacks belonging to dead threads are freed by the
979 next exiting thread, no need to gc to collect thread stacks anymore
980 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
981 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
982 ** bug fix: dynamic variable and thread start related gc lossage
983 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
985 * fixed some bugs revealed by Paul Dietz' test suite:
986 ** SUBTYPEP is slightly more accurate on heinously complicated
987 CONS types where some of the members have uncertain (in the
988 NIL, NIL sense) type relationships to each other.
990 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
991 no longer purified unless :PURIFY T is explicitly specified.
992 ** Non-purified cores are significantly smaller than before
994 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
995 * new feature: timers based on Zach Beane's excellent timer package
996 * added support for the following external formats: koi8-u,
997 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
998 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
999 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1000 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1001 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1002 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1003 * incompatible change: a threaded SBCL will no longer revert to
1004 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1005 * bug fix: interrupts are disabled until startup is complete; no
1006 more sigsegvs when receiving a signal to soon
1007 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1009 * bug fix: add a workaround for the memory randomization features in
1010 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1011 workaround will only be in effect on systems with the proc filesystem
1013 * bug fix: printing objects of type HASH-TABLE signals a
1014 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1016 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1017 have been running for over 50 days. (reported by Gilbert Baumann)
1018 * bug fix: the logic for getting names of functions gets less
1019 confused when confronded with alternate-metaclass
1020 funcallable-instances. (reported by Cyrus Harmon)
1021 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1022 consistent, even on internal alternate-metaclass objects.
1023 * bug fix: flush closure information collected by physical
1024 environment analysis prepass before the main pass. (bug reported
1026 * bug fix: compiler pack phase does not modify a hashtable, which is
1027 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1028 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1029 may be expanded. (reported by Alexander <malishev> on c.l.l)
1030 * bug fix: no more style-warnings when compiling code using LAST
1031 under high-SPEED low-SPACE settings. (reported by David Wallin
1033 * fixed some incompatibilities between SBCL's MOP and the MOP
1035 ** the METAOBJECT class is now implemented;
1036 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1037 STANDARD-OBJECT, as required;
1038 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1039 now compatible (as required by VALIDATE-SUPERCLASS); there
1040 remains a consistency requirement over the presence or absence
1041 of the FUNCTION class in the superclass list of finalized
1042 classes; see the manual for more details;
1043 ** the :AROUND method for COMPUTE-SLOTS on
1044 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1045 requested slot ordering.
1047 ** bug fix: parent thread now can be gc'ed even with a live
1049 ** bug fix: binding a special with PROGV to no value is not
1050 the same as that symbol not having been bound (thanks to
1052 ** bug fix: binding specials is thread safe (thanks to
1054 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1055 and SB-SPROF do not die with 'no handler for signal XX in
1056 interrupt_handle_now(..)' anymore
1057 ** bug fix: WITH-TIMEOUT works with multiple running threads
1058 * fixed some bugs revealed by Paul Dietz' test suite:
1059 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1060 the :method-class keyword argument.
1062 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1063 * new port: the Solaris operating system on x86 processors is now
1064 mostly supported, though some rough edges in the environment
1065 remain. (thanks to Daisuke Homma)
1066 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1067 space for its heap. (thanks to Thiemo Seufer)
1068 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1069 closures. (thanks to Thiemo Seufer)
1070 * minor incompatible change: eof selects abort in the debugger.
1071 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1072 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1074 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1075 deprecated, and will go away in a future revision of SBCL.
1076 * minor incompatible change: GC-ON and GC-OFF are no longer
1077 implemented with a counter, it does not matter how many times gc
1078 is switched on or off
1079 * bug fix: discriminating functions for generic function classes
1080 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1081 make invalid assumptions about method precedence order. (reported
1083 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1084 platforms. (thanks to Thiemo Seufer)
1085 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1086 * bug fix: correct pathname printing with printer escaping is on.
1087 (thanks to Kevin Reid)
1088 * bug fix: complex VOP definitions in "user-space" no longer trigger
1089 package locks. (reported by Zach Beane)
1090 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1091 for funcallable-instance functions, and is no different from
1093 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1094 characters in the current readtable. (reported by Nicholas Neuss)
1095 * bug fix: SUBTYPEP on various CONS types returns more a more
1096 accurate acknowledgment of its certainty.
1097 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1098 lists in some cases. This partially fixes bug 384.
1099 * flush all standard streams before prompting in the REPL and the
1101 * bug fix: signal handling and triggering gc do not conflict
1102 directly anymore, in particular a high frequency sb-sprof does
1103 not prevent gc from running
1104 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1105 approximation for timezone and DST information between the
1106 universal time epoch and the smallest negative 32-bit time_t.
1107 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1108 year 1899 so long as the encoded time is non-negative.
1109 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1110 the correct answer for non-constant rotate amounts when performing
1111 an inline 32-bit rotation.
1113 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1114 there is only one thread in the session
1115 ** bug fix: memory leak for streams created in one thread and
1116 written to in another
1117 ** bug fix: lockup when compiled with gcc4
1118 ** bug fix: race that allows the gc to be triggered when gc is
1120 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1121 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1123 ** bug fix: finished threads are gc'ed properly
1124 * fixed some bugs revealed by Paul Dietz' test suite:
1125 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1126 has arguments with a different set of applicable methods from
1127 the orignal arguments.
1128 ** The type error thrown by MAP now has a correct expected-type
1130 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1131 name a compiled function.
1132 ** Three MISC tests where a large bignum was improperly coerced to
1133 a float that couldn't represent that bignum during type
1134 derivation were fixed.
1135 ** SUBTYPEP can now handle types involving the negation of a
1136 list-form FUNCTION type.
1137 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1138 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1139 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1141 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1142 * New feature: Experimental support for bivalent streams: streams
1143 opened with :element-type :default now allow character and binary
1144 (unsigned-byte 8) I/O
1145 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1146 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1148 * bug fix: no more highly sporadic "couldn't check whether ~S is
1149 readable" when reading a stream and an interrupt hits in the middle
1150 of a select system call
1151 * compiler better recognizes complex arrays (reported by Eduardo
1153 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1154 for variables on SBCL built with linkage-tables. (reported by Luis
1156 * various error reporting improvements.
1157 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1158 (Thanks to Thiemo Seufer)
1159 * bug fix: nested reader invokations work correctly
1160 * bug fix: it is possible to have simultaneous references to foreign
1161 code and foreign data with the same name.
1163 ** added x86-64 support
1164 ** incompatible change: the threading api now works with thread
1165 objects instead of thread ids
1166 ** bug fix: threads are protected from signals and interruption when
1167 starting up or going down
1168 ** bug fix: a race where an exiting thread could lose its stack to gc
1169 ** bug fix: don't halt on infinite error in threads if possible
1170 ** fixed numerous gc deadlocks introduced in the pthread merge
1171 ** bug fix: fixed thread safety issues in read and print
1172 ** bug fix: debugger doesn't hang on session lock if interrupted at
1173 an inappropriate moment
1174 ** bug fix: run-program is now thread safe(r)
1175 ** bug fix: inner with-recursive-lock no longer releases the mutex
1176 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1177 * fixed some bugs revealed by Paul Dietz' test suite:
1178 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1179 EXPECTED-TYPE slots filled.
1180 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1182 ** MAKE-CONDITION accepts classes as type-designators.
1183 ** COMPILE may never return NIL.
1184 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1185 range before calling Unix time functions
1187 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1188 * numerous signal handling fixes to increase stability
1189 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1190 * minor incompatible change: we now correctly canonize default
1191 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1192 as per the MOP, rather than the historical (INITARG INITFUNCTION
1193 INITFORM). (reported by Bruno Haible)
1194 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1195 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1196 for more information.
1197 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1198 pathname is a directory pathname.
1199 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1200 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1202 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1203 :METACLASS argument in addition to a class name. (reported by
1204 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1205 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1206 (reported by Bruno Haible; patch from Gerd Moellmann)
1207 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1209 * bug fix: more cleanups to the floating point exception handling on
1210 x86-64 (thanks to James Knight)
1211 * bug fix: the compiler does not try to constant fold calls of
1212 COERCE and BIT-* functions when they return freshly constructed
1213 objects. (reported by Nikodemus Siivola and Paul Dietz)
1214 * optimization: improved the MIPS versions of generic (in the
1215 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1216 * optimization: direct conversion of (unsigned-byte 32) to floats on
1217 the PowerPC platform.
1218 * optimization: structure instances with raw slots now use less
1219 memory, and probably show better memory locality. (thanks to
1221 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1222 are treated as requests for stack allocation on the x86-64,
1223 Alpha-32, PPC and SPARC platforms.
1224 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1225 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1226 (thanks to Tony Martinez)
1228 ** gcing a dead thread can no longer lead to lockups
1229 ** threads block signals until they are set up properly
1230 ** errno is no longer shared by threads
1231 ** interrupt-thread restores the eflags register on x86
1232 ** fixed some lockups due to gc/thread interaction
1233 * fixed some bugs revealed by Paul Dietz' test suite:
1234 ** invalid dotted lists no longer raise a read error when
1235 *READ-SUPPRESS* is T
1236 ** signal an error if a symbol that names a declaration is used
1237 as the name of a type, or vice versa
1238 ** allow using the (declare (typespec var*)) abbreviation for
1239 (declare (type typespec var*)) with all type specifiers
1240 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1241 ** always return NIL from PPRINT-POP when OBJECT is NIL
1242 ** don't signal errors when pretty-printing malformed LABELS,
1243 FLET or MACROLET forms
1244 ** declarations in a DOLIST body are also in scope for the
1246 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1248 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1251 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1252 * fixed cross-compiler leakages that prevented building a 32-bit
1253 target with a 64-bit host compiler.
1254 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1255 opened with :IF-EXISTS :OVERWRITE.
1256 * fixed bug 281: error for an invalid qualifier in a short-form method
1257 combination method is not signalled until the faulty method is called.
1258 * bug fix: iteration variable type inferrer failed to deal with open
1259 intervals. (reported by Alan Shields)
1260 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1261 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1262 or more runtime options were provided to the sbcl binary.
1263 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1264 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1265 * optimization: there's now a fast-path for fixnum arguments in the
1266 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1267 * optimization: the code generated on x86-64 is more compact thanks
1268 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1269 * bug fix: floating-point exception handling now partly works on
1270 x86-64. (Thanks to James Knight)
1271 * improvement to the MIPS backend from Thiemo Seufer: C-style
1272 64-bit long long arguments and return values to alien functions
1274 * fixed some bugs revealed by Paul Dietz' test suite:
1275 ** the type-error signalled from WARN has a filled-in DATUM slot.
1276 ** the type-error required when a stream is not associated with
1277 a file has the stream as its datum.
1278 ** type-errors on single-floats on x86-64 no longer have
1279 :INVALID-OBJECT as the datum
1280 ** the type-errors signalled for invalid function names now have
1281 a correct expected type
1282 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1283 for structures defined with a :TYPE. Documentation strings for
1284 typed structures are no longer immediately discarded
1285 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1286 broadcast streams and synonym streams. FILE-LENGTH now also works
1287 on broadcast streams.
1289 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1290 * incompatible change: the --noprogrammer option, deprecated since
1291 version 0.7.5, has been removed. Please use the equivalent
1292 --disable-debugger option instead.
1293 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1295 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1296 inoperational for a while now) has been completely removed.
1297 * null lexical environments are now printed as #<NULL-LEXENV>,
1298 significantly reducing the amount of clutter in typical backtraces.
1299 * documentation on weak pointers, finalization, and after GC hooks
1300 has been added to the manual.
1301 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1302 as other specialized array types, is much faster. SUBSEQ and
1303 COPY-SEQ on such arrays have also been sped up.
1304 * optimization: EQL is now more efficient when at least other argument
1305 is known to be of type (OR FIXNUM (NOT NUMBER)).
1306 * fixed bug: the runtime is now less vulnerable to changes in the
1307 size of the SBCL object on OS X, and virtual memory is reserved for
1308 all spaces that need to be at a fixed address.
1309 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1310 * fixed bug: finalizers and after GC hooks that cause consing are now
1312 * fixed bug: compiler error messages and summaries are now printed to
1313 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1314 * fixed inference of the upper bound of an iteration variable.
1315 (reported by Rajat Datta).
1316 * fixed bug 211e: calling local functions with duplicated constant
1317 keyword argument no longer causes a bogus style warning about an
1319 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1320 ftype declaration to be disregarded. (reported by Dave Roberts)
1321 * fixed bug 373: caused by erronous compilation of references to alien
1322 variables in the runtime on ppc/darwin.
1323 * fixed bug 376: CONJUGATE type deriver.
1324 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1325 (reported by Baughn on #lisp)
1326 * a cross-compiler bug on non-x86 platforms has been identified and
1327 fixed. (thanks to Bruno Haible)
1328 * improvements to the MIPS runtime code for increased stability.
1329 (thanks to Thiemo Seufer)
1330 * increased the maximimum compact environment size to allow
1331 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1332 * partial workaround for bug 135: don't name the function we're
1333 calling for hairy cases of EVAL, so as not to accumulate one environment
1334 entry per call to EVAL. (reported by Kevin Reid)
1335 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1336 * optimization: type testing for non-vector arrays should be faster.
1337 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1339 * bug fix: setting 31st element of a bit vector to zero did not work
1341 * bug fix: redefining a class definition which failed due to a
1342 previous accessor / function clash now works (but see BUGS entry
1343 #380 for more problems in this area). (thanks to Zach Beane)
1344 * the long form of DEFINE-METHOD-COMBINATION disables method group
1345 checking when given a single method group with pattern *. (thanks
1347 * on x86 compiler supports stack allocation of results of simple
1348 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1349 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1350 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1351 * fixed some bugs related to Unicode integration:
1352 ** the restarts for recovering from input and output encoding
1353 errors only appear when there is in fact such an error to
1355 * fixed some bugs revealed by Paul Dietz' test suite:
1356 ** MISC.549 and similar: late transformation of unsafe type
1357 assertions into derived types caused unexpected code
1359 ** SCALE-FLOAT type deriver is less wrong.
1360 ** type derivers for EXP, LOG and similar functions compute result
1361 types for complex arguments better.
1362 ** (MISC.563) CONJUGATE type deriver works for very restricted
1364 ** out-of-line type testers for character strings are available.
1365 ** EQUAL compiler transform understands specialness of objects
1367 ** accessing double-floats stored on the stack now works on x86-64.
1368 ** debugger internals could sometimes create invalid lispobjs,
1369 resulting in GC crashes.
1370 ** MISC.548: type check weakening can convert required type into
1372 ** initialization forms for bindings are not in scope of free special
1375 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1376 * incompatible change: thread support for non-NPTL systems has
1377 been removed - locking is buggy and unreliable. A threaded
1378 SBCL build will now warn at startup and refuse to create
1379 new threads, unless futex support is detected in the kernel
1380 * incompatible change: the top level REPL now has only an ABORT
1381 restart associated with it, not TOPLEVEL and ABORT as it used to.
1382 TOP and TOPLEVEL are now available as debugger commands for
1383 returning to the top level.
1384 * incompatible change: forms evaluated in the REPL now use the
1385 global optimization policy.
1386 * incompatible change: user- and system-initialization files are
1387 no longer processed with LOAD, but by READ and EVAL; hence the
1388 global optimization policy, startup package, readtable, etc,
1390 * The .fasl file format number has been incremented because of
1391 various incompatible changes.
1392 * internal entry point details and argument counts no longer appear
1393 in backtraces unless explicitly requested by setting
1394 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1395 * built-in and standard functions no longer have names like "top
1396 level local call to FOO".
1397 * fixed bug 32: functions defined in non-null lexical environments
1398 now have more legible printed representation
1399 * fixed bug 33: functions defined in non-null lexical environemnts
1400 are now more amenable to inspection by INSPECT.
1401 * workaround for bug 354: XEPs no longer appear in backtraces unless
1402 explicitly requested.
1403 * fixed bug: receiving the signal which results from attempting to
1404 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1405 SIGBUS on other *BSDs) on architectures where the C stack is also
1406 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1407 to debugger confusion or wild instability. (reported by Cheuksan
1409 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1410 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1411 (reported by Lutz Euler)
1412 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1413 compiled code. (reported by Paul Dietz)
1414 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1415 specialized parameter to a method to something that is not TYPEP
1416 the specializer is now possible.
1417 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1418 face of package deletion.
1419 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1420 pathname. (reported by David Tolpin on c.l.l)
1421 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1422 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1423 STRING-STREAM and FILE-STREAM, respectively.
1424 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1425 than previously; in addition, its test suite is now run on build.
1426 (thanks to Luke Gorrie)
1427 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1428 * optimization: added a immediate representation for single-floats
1430 * fixed some bugs revealed by Paul Dietz' test suite:
1431 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1432 ** attempting to create a package with a colliding nickname causes
1433 correctable errors to be signalled.
1434 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1435 ** several x86-64 backend bugs related to sign-extension of immediate
1438 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1439 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1440 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1442 * fixed loading of multiply forward-referenced layouts.
1443 (thanks to Cheuksan Wang)
1444 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1445 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1446 (see COMPILE-FILE documentation for details of :PRINT option)
1447 * fixed bugs 19 and 317: fixed-format floating point printing is
1448 more accurate. This also fixes a bug reported by Adam Warner
1449 related to the ~@F format directive.
1450 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1452 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1453 dispatch table if the from-char is a dispatch macro character.
1454 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1455 by the wordsize no longer ignore the last word. (reported by Lutz
1457 * fixed bug in type checking of dynamic-extent variables. (reported
1459 * optimization: sequence traversal functions use their freedom to
1460 coerce function designators to functions.
1461 * optimization: code with many calls to CLOS methods specialized on
1462 CLOS classes has had redundant type checks removed.
1463 * optimization: added declarations to speed up operations that access
1464 the internal character database (for example STRING-UPCASE)
1465 * optimization: comparison operations between floats and sufficiently small
1466 fixnums no longer create extra rationals
1467 * fixed some bugs related to Unicode integration:
1468 ** portions of multibyte characters at the end of buffers for
1469 character-based file input are correctly transferred to the
1470 start of the buffer at the next read.
1471 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1472 passing it through to OPEN.
1473 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1474 argument given to internal calls to OPEN.
1475 ** the built-in comment readers (introduced by character sequences
1476 ";" and "#|") are more forgiving to encoding errors; they will
1477 STYLE-WARN and then attempt to resync the stream at a character
1478 boundary. (thanks to Teemu Kalvas)
1479 * fixed some bugs in the x86-64 port:
1480 ** Negative short int return values from c-calls are sign-extended
1482 ** The stack is aligned to 16-bytes for c-calls, as required by
1483 the ABI. (thanks to Cheuksan Wang)
1484 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1485 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1486 for enums. (thanks to Vincent Arkesteijn)
1487 ** Multiple small optimizations and bugfixes for floating point
1489 * fixed some bugs revealed by Paul Dietz' test suite:
1490 ** Space, Tab, Linefeed, Return and Page have the invalid
1491 secondary constituent character trait.
1492 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1494 ** WITH-INPUT-FROM-STRING should only update the index place on
1496 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1497 works more reliably.
1498 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1499 with one-dimensional arrays with fill pointers.
1500 ** TYPE-OF failed on a complex with an integer realpart and a
1502 ** compiler failure during type inference for the code of form
1503 (IF (EQL X (THE ...)) ...) (MISC.535).
1505 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1506 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1507 platform. The port supports 61-bit fixnums, large memory spaces
1508 and reloading shared object files.
1509 * enhancement: saving cores with foreign code loaded is now
1510 supported on ppc/Darwin in addition to the previously supported
1512 * enhancement: the statistical profiler now walks deeper into the
1513 call stack for more meaningful call-graphs and accrued time
1514 reports (x86/x86-64 only). It also now reports time spent in
1516 * enhancement: it is now possible to trace most individual methods
1517 of a generic function in addition to tracing the generic function
1519 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1521 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1522 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1523 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1524 produces an error. (thanks to Vincent Arkesteijn)
1525 * bug fix: NAMESTRING on pathnames with :WILD components in their
1526 directories works correctly. (thanks to Artem V. Andreev)
1527 * fixed bug 125: compiler preserves identity of closures. (reported
1529 * bug fixed: functions with &REST arguments sometimes failed with
1530 "Undefined function" when compiled with (DEBUG 3). (reported by
1532 * bug fix: overflow during compiling of setting element of a bit
1533 vector with constant index and value. (reported by Timmy Douglas)
1534 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1535 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1536 * on x86 compiler partially supports stack allocation of dynamic-extent
1538 * GO and RETURN-FROM do not check the extent of their exit points
1539 when compiled with SAFETY 0.
1540 * fixed some bugs related to Unicode integration:
1541 ** encoding and decoding errors are now much more robustly
1542 handled; it should now be possible to recover even from invalid
1543 input or output to the terminal. (thanks to Teemu Kalvas)
1544 ** provided a first cut at implementing STRING-TO-OCTETS and
1545 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1546 ** altered the SB-MD5 contributed module slightly, changing the
1547 interface just enough for it to be supportable for builds where
1548 lisp characters are not eight bits.
1549 * fixed some bugs revealed by Paul Dietz' test suite:
1550 ** the FORMATTER-generated functions for ~V[ conditionals require
1551 the correct number of arguments.
1552 ** READ-FROM-STRING returns the mandated second value when applied
1553 to displaced strings.
1554 ** the #\Rubout and #\Backspace characters are treated as invalid
1555 constituent characters by the tokenizer.
1557 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1558 * new feature: reloading changed shared object files with
1559 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1560 * new feature: references to foreign variables and functions
1561 can now be compiled and loaded before the corresponding shared
1562 object file is loaded, as long as the foreign definitions are
1563 available at runtime.
1564 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1565 supported. (thanks to Dan Debertin)
1566 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1567 just keywords. (thanks to Vincent Arkesteijn)
1568 * fixed bug #331: structure-class instances corresponding to
1569 DEFSTRUCT forms are now created eagerly.
1570 * fixed bug #345: backtraces from calls to undefined functions work
1571 on x86 as well. Related bug #61 is now also partially fixed on x86
1572 (backtraces from throws to unknown catch tags.)
1573 * bug fix: lambda-list parsing is now stricter vrt. order and number
1574 of lambda-list keywords.
1575 * bug fix: as specified by AMOP, an error is signalled if a
1576 class-option appears multiple times in a DEFCLASS form. (reported
1578 * bug fix: RANDOM can be compiled when the compiler derives the type
1579 of its numeric argument as a disjoint set of small integers.
1580 (reported by Paul Dietz)
1581 * bug fix: starting a core saved with shared objects loaded when
1582 those objects are not available no longer causes threaded SBCL to
1583 hang. (reported by Sean Ross)
1584 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1586 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1587 argument, which must always be NIL. (reported by Kalle Niemitalo)
1588 * bug fix: printing 1.0d+23 no longer results in an error.
1589 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1590 * bug fix: structure slot setters preserve evaluation order. (thanks
1592 * fixed some bugs related to Unicode integration:
1593 ** RUN-PROGRAM can allow its child to take input from a Lisp
1594 stream. (reported by Stefan Scholl)
1595 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1596 locales. (reported by Ken Causey)
1597 * fixed some bugs revealed by Paul Dietz' test suite:
1598 ** INCF, DECF and REMF evaluate their place form as specified in
1600 ** FORMATTER expands ~{ ~} iteration directives with V or #
1601 parameters correctly.
1602 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1603 consequent uses no arguments correctly.
1604 ** the system has a partial understanding of the (COMPLEX RATIO)
1606 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1609 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1610 * new feature: a build-time option (controlled by the :SB-UNICODE
1611 keyword feature, enabled by default) for building the system with
1612 support for the entire 21-bit character space defined by the
1614 * new feature: the system now has rudimentary external-format
1615 support; the primary user-visible change at this time is that
1616 characters with the high bit set (such as lower-case-e-acute) will
1617 print correctly to a terminal in a UTF-8 environment.
1618 * minor incompatible change: BASE-CHAR no longer names a class;
1619 however, CHARACTER continues to do so, as required by ANSI.
1620 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1621 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1622 has been moved to the SB-EXT package (temporarily re-exported from
1624 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1625 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1627 * fixed bug #351: better error-handlind and reporting for malformed
1629 * fixed bug #350: bignum-printing is now more memory-efficient,
1630 allowing printing of very large bignums, eg. (expt 2 10000000).
1631 (reported by Bruno Haible)
1632 * fixed bug #302: better primitive-type selection for intersection
1634 * fixed bug #308: non-graphic characters now all have names, as
1635 required. (reported by Bruno Haible)
1636 * bug fix: redefining a class with different superclasses now correctly
1637 removes it from the direct-subclasses of its previous superclasses.
1638 (reported by David Morse)
1639 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1640 new value now works. (reported by Bruno Haible)
1641 * bug fix: correct canonicalization of multiple non-standard slot
1642 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1643 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1644 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1645 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1646 now exists, an signals an error.
1647 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1648 during macroexpansion for non-iteration variables as well. (reported
1649 by Bruno Haible for CMUCL)
1650 * bug fix: Cyclic structures and unprintable objects in compiler
1651 messages no longer cause errors. (reported by Bruno Haible)
1652 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1653 and READ-FROM-STRING all now return a primary value of NIL if
1654 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1655 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1656 (reported by Bruno Haible for CMUCL)
1657 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1658 do not designate a condition. (reported by Bruno Haible for
1660 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1661 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1662 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1663 permitting resolution in favour of any of the conflicting symbols.
1664 (reported by Bruno Haible for CMUCL)
1665 * bug fix: EQUAL compiler optimizations is less aggressive on
1666 strings which can potentially compare true despite having distinct
1667 specialized array element types.
1668 * bug fix: unit enumerations can be defined without dividing by
1669 zero. (thanks to Vincent Arkesteijn)
1670 * FORMAT compile-time argument count checking has been enhanced.
1671 (report from Bruno Haible for CMUCL)
1672 * a partial workaround for the bug 262: the compiler does not try to
1673 inline-expand a local function doing RETURN-FROM from a deleted
1674 BLOCK. (thanks to Peter Denno for the bug report and to David
1675 Wragg for the simple test case)
1676 * fixed some bugs revealed by Paul Dietz' test suite:
1677 ** RENAME-PACKAGE allows all package designators as new package
1679 ** constraint propagation and lambda variable substitution are
1680 more cautious in dealing with partially deleted code.
1681 ** compiler sometimes lost reoptimization passes.
1682 ** CERROR, given a condition as condition designator, passes its
1683 remaining arguments to the continue format control without
1685 ** Case-altering FORMAT directives work correctly on non-ASCII
1687 ** The REAL type specifier handles bounds outside the floating
1688 point ranges without signalling FLOATING-POINT-OVERFLOW.
1689 ** Functions with IR1-transformations can create intercomponent
1690 references to global functions.
1691 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1693 ** FORMAT ~R treats a nil value for its first parameter correctly.
1695 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1696 * enhancement: saving cores with foreign code loaded is now
1697 supported on x86/NetBSD and sparc/Linux in addition to the previously
1698 supported platforms.
1699 * bug fix: on some platforms repeated installations caused multiple
1700 copies of HTML documentation to be installed -- should not happen
1701 any more. (reported by Stefan Scholl)
1702 * bug fix: parsing self-recursive alien record types multiple times
1703 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1704 original patch by Helmut Eller for CMUCL)
1705 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1706 (thanks to Richard Kreuter)
1707 * bug fix: defining classes whose accessors are methods on existing
1708 generic functions in other (locked) packages no longer signals
1709 bogus package lock violations. (reported by François-René Rideau)
1710 * bug fix: special variables as DEFMETHOD parameters no longer have
1711 associated bogus type declarations. (reported by David Wragg and
1713 * bug fix: read-write consistency on streams of element-type
1714 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1715 * bug fix: redefiniton of the only method of a generic function with
1716 no DEFGENERIC no longer emits a full WARNING. In addition,
1717 redefinition of generic functions with no DEFGENERIC to an
1718 incompatible lambda list now signals an error. (thanks to Zach
1720 * bug fix: DEFGENERIC now works even when there's a function of the
1721 same name in an enclosing lexical environment. (thanks to Zach
1723 * fixed compiler failure, caused by instrumenting code during
1724 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1725 * optimization: added loop analysis and improved register allocation
1726 * fixed some bugs revealed by Paul Dietz' test suite:
1727 ** POSITION on displaced vectors with non-zero displacement
1728 returns the right answer.
1729 ** (SIMPLE-STRING) is a valid type specifier for sequence
1731 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1733 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1734 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1736 ** The justification version of the FORMAT ~< directive treats
1737 non-zero minpad parameter correctly.
1739 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1740 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1741 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1742 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1743 the supported interface.
1744 * new feature: Single-stepping of code compiled with DEBUG 2 or
1745 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1746 * new feature: saving cores with foreign code loaded is now
1747 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1748 Timothy Moore's work for CMUCL)
1749 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1750 parameters to * instead of NIL if no initform is supplied.
1751 (reported by Johan Bockgaard)
1752 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1753 correctly when FUNCALL appears as the car of the form. Note:
1754 despite this FUNCALL forms are not currently subject to
1755 compiler-macro expansion. (port of Raymond Toy's fix for the
1756 same from CMUCL, reported by Johan Bockgaard)
1757 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1758 (thanks for Teemu Kalvas)
1759 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1760 case instead of terminating the iteration (thanks for Julian
1761 Squires, Sean Champ and Raymond Toy)
1762 * bug fix: incorrect expansion of defgeneric that caused a style
1763 warning. (thanks for Zach Beane)
1764 * bug fix: slot accessor effective method computation works properly
1765 for classes with multiple non-standard applicable methods on
1766 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1767 * on x86 compiler supports stack allocation of results of LIST and
1768 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1769 CMUCL implementation by Gerd Moellmann)
1770 * fixed some bugs revealed by Paul Dietz' test suite:
1771 ** FORMAT strings with both the ~<~:;~> form of the justification
1772 directive and pretty-printing directives cause an error.
1774 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1775 * incompatible change: the internal functions
1776 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1777 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1778 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1779 instead of the old functions.
1780 * new feature: on platforms where "dladdr" is available foreign
1781 function names now appear in backtraces. (based on Helmut Eller's
1783 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1784 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1786 * documentation: ASDF manual is now distributed with SBCL.
1787 * bug fix: SBCL can now load its contributed modules with REQUIRE
1788 even if the system-provided entries have been removed from
1789 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1791 * bug fix: backtraces involving undefined functions or assembly
1792 routines are more informative. (thanks to Brian Downing)
1793 * bug fix: mutually referent alien structures now work correctly.
1794 (reported by Rick Taube)
1795 * bug fix: structures defined by WITH-ALIEN can be referred to
1796 within other definitions in the same WITH-ALIEN.
1797 * bug fix: division operators (MOD, TRUNCATE and the like) with
1798 constant zero divisors and integer dividends no longer generate
1800 * bug fix: provide default methods for INPUT-STREAM-P and
1801 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1802 * bug fix: improve the detection and resolution of MOP metacycles
1803 (where effective method computation depends on the generic function
1804 having its effective method computed). (reported by Bruno Haible)
1805 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1806 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1807 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1808 * bug fix: pathnames with relative directory components can be
1809 represented relative to default pathnames.
1810 * optimization: in taking the GCD of bignums, reduce the two bignums
1811 to approximately the same size (using Euclid's algorithm) before
1812 applying the more sophisticated binary GCD. (thanks to Juho
1814 * optimization: COUNT on bitvectors now operates word-at-a-time.
1815 * optimization: ASH with a positive, but not necessarily constant,
1816 (leftwards) shift, when in a modular context, is compiled to a
1818 * fixed some bugs revealed by Paul Dietz' test suite:
1819 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1820 if the corresponding argument is NIL.
1821 ** FORMAT directives accepting a minpad parameter treat negative
1823 ** literal commas as character parameters to FORMAT directives are
1825 ** literal spaces directly after ~<Newline> directives within a
1826 format-logical-block (~:< ~@:>) do not induce :FILL-style
1827 conditional newlines.
1828 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1829 as specified: it no longer includes conditional newlines).
1830 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1832 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1833 :PER-LINE-PREFIX argument does not evaluate to a string.
1835 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1836 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1837 the manual for details; for now, package locks can be disabled by
1838 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1839 if no major problems are found then it is likely that they will be
1840 compiled in unconditionally.
1841 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1842 unsupported operators on all platforms. To load a shared library
1843 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1844 object file, link it into a shared library outside of SBCL and
1845 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1847 * bug workaround: ROOM T and the GC were not working together
1848 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1849 reporting facility aren't preserved. That reporting has been
1850 disabled, so now until and unless someone figures out how to make it
1851 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1852 * minor incompatible change: as threatened around sbcl-0.8.0, the
1853 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1854 The symbols are also exported from SB-PCL for backwards
1855 compatibility, but more so than before SB-PCL should be treated as
1856 an implementation-internal package.
1857 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1859 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1860 It is known as of this release not to work on the Alpha, however.
1861 * fixed bug #167: errors signalled due to illegal syntax in method
1862 bodies are now more legible.
1863 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1864 designators and can hence be used with TYPEP.
1865 * fixed bug #333: CHECK-TYPE now ensures that the type error
1866 signalled, if any, has the right object to be accessed by
1867 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1868 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1869 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1871 * fixed bug #334: programmatic addition of slots using specialized
1872 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1873 and :ALLOCATION :CLASS slots.
1874 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1875 float exponents. (rereported by Peter Seibel)
1876 * fixed a bug: #\Space (and other whitespace characters) are no
1877 longer considered to be macro characters in standard syntax by
1878 GET-MACRO-CHARACTER.
1879 * fixed bug: initialization of condition class metaobjects no longer
1880 causes an instance of the condition to be created. (reported by
1882 * fixed bug: it is now possible to have slots such that
1883 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1884 description is neither :INSTANCE nor :CLASS.
1885 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1886 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1887 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1888 non-standard SLOT-DEFINITION-ALLOCATION values.
1889 * fixed bugs: various race conditions handling exiting threads.
1890 CL-PPCRE's thread torture test now passes.
1891 * fixed arguably-a-bug: GC time proportional to number of threads in
1892 system even when most of them are idle
1893 * optimization: improved performance of BIT and SBIT on bit-vectors.
1894 * .fasl file incompatibility: The fasl file version number has
1895 been incremented because of changes associated with package locks.
1897 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1898 * minor incompatible change: the system no longer provides
1899 optimization hints (conditions which are TYPEP
1900 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1901 mode; these hints are still emitted when higher SPEED optimization
1903 * new contrib module: a sampling profiler (profiling by statistical
1904 sampling, rather than by instrumenting functions) is available as
1905 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1906 * the behaviour of the standard function ED is now customizeable by
1907 third parties through a hook variable: see ED's documentation
1908 string for information on the protocol.
1909 * the compiler no longer emits efficiency notes for (FUNCALL X)
1910 when the type of X is uncertain under default optimization
1912 * fixed bug 276: mutating a binding of a specialized parameter to a
1913 method to something that is not TYPEP the specializer is now
1915 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1916 exhibit the required behaviour on the x86 platform. (thanks to
1917 Peter van Eynde, Eric Marsden and Bruno Haible)
1918 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1919 even for difficult arguments. (reported by Peter Graves)
1920 * fixed bug 141a: the backquote printer now descends quoted
1922 * fixed another bug in backquote printing: no more destructive
1923 modification of the form's list structure. (reported by Brian
1925 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1926 move between its address being taken and the call to
1927 interrupt_thread, fixing a crashing race condition.
1928 * the SB-POSIX contrib implementation has been adjusted so that it
1929 no longer exhibits ridiculously poor performance when constructing
1930 instances corresponding to C structs.
1932 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1933 * minor incompatible change: the sb-grovel contrib now treats C
1934 structures as alien (in the sense of SB-ALIEN) objects rather than
1935 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1936 has implications for memory management of client code
1937 (sb-grovel-returned objects must now be manually managed) and for
1938 type safety (alien objects now have full types).
1939 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1940 used to control emission of compiler diagnostics, rather than the
1941 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1942 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1943 quality should be considered deprecated.
1944 * install.sh now installs the user manual as well
1945 * (not quite a new documentable feature, but worth considering in
1946 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1947 beginnings of a semantically meaningful condition hierarchy is
1948 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1950 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1951 designator as the defaults argument.
1952 * fixed bug: Displaced arrays whose displaced-to array has become
1953 too small now cause ARRAY-DIMENSION to signal an error, providing
1954 for safer bounds-checking. (reported by Bruno Haible)
1955 * fixed bug: DEFCLASS slot definitions with identical :READER and
1956 :WRITER names now signal a reasonable error. (reported by Thomas
1958 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1960 * fixed bug: Invalid slot specification errors now print correctly.
1961 (thanks to Zach Beane)
1962 * fixed bug 320: Shared to local slot value transfers in class
1963 redefinitions now happen corrently from superclasses as
1964 well. (reported by Bruno Haible)
1965 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1967 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1968 improper lists correctly. (reported by Bruno Haible)
1969 * fixed bug 313: source-transform for <fun-name> was erroneously
1970 applied to a call of a value of a variable with name <fun-name>.
1971 (reported by Antonio Menezes Leitao)
1972 * fixed bug 307: The obsolete instance protocol ensures that
1973 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1974 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1975 remove dead unknown-values globs from the middle of the stack.
1976 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1977 easier to resume long computations after using *BREAK-ON-SIGNALS*
1978 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1979 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1980 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1981 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1982 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1983 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1985 * fixed a (fairly theoretical) bug in string printing: if
1986 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1987 does not have array-element-type equal to the most general string
1989 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1990 function name. (thanks to Nikodemus Siivola)
1991 * fixed bug: compile-time format string checker failed on
1992 non-closed ~{. (reported by Thomas F Burdick)
1993 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1994 don't warn on legal loop code involving more than one aggregate
1995 boolean. (thanks to Nikodemus Siivola)
1996 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1997 in keyword position to MAKE-PACKAGE was not being reported as
1999 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2000 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2002 * fixed bugs 280 and 312: the checking for multiple definitions in a
2003 file is less likely to become confused by uses of inline
2005 * fixed bug: the #S reader macro performs the keyword coercion
2006 specified for slot names. (reported by Kalle Niemitalo)
2007 * fixed bug: lambda lists may contain symbols whose names start with
2008 & but are not lambda-list-keywords; their occurrence triggers a
2010 * fixed bug 321: define-method-combination argument lambda lists do
2011 not cause invalid code to be generated when &optional or &aux
2012 variables have default values. (reported by Bruno Haible)
2013 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2014 have CLOS classes; this appears also to have fixed reported
2015 instability in stack exhaustion detection.
2016 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2017 MUFFLE-WARNING when no associated restart is present is now
2019 * optimization: rearranged the expansion of various defining macros
2020 so that each expands into only one top-level form in a
2021 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2023 * optimization: used a previously little-used slot in symbols to
2024 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2026 * fixed some bugs revealed by Paul Dietz' test suite:
2027 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2029 ** Fixed an optimization bug related to inheritance of initforms
2030 from local to shared slots.
2031 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2032 ** CAST optimizer forgot to flush argument derived type.
2033 ** print/read consistency on floats is now orders of magnitude
2034 more likely. (thanks also to Bruno Haible for a similar report
2036 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2037 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2038 for objects of type REAL. Make it so.
2039 ** SXHASH is specified (infelicitously) to respect similarity,
2040 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2041 it so. (thanks to Markus Ziegler)
2042 ** on the Alpha, the compiler succeeds in compiling functions
2043 returning a known number of arguments greater than 63.
2044 ** fixed handling of invalid NIL arguments in keyword position
2046 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2048 ** provided a readably-printable representation for RANDOM-STATE
2050 ** ensured that pathnames, where they have a namestring, always
2051 print using #P"..." syntax.
2053 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2054 * Support for the forthcoming 2.0 version of the NetBSD kernel
2055 running on x86 hardware has been added. (thanks to Perry
2056 E. Metzger most immediately, and others for their past work)
2057 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2058 current release version; SBCL's OpenBSD support had been broken
2059 since about the time of OpenBSD's switch to ELF binary format.)
2060 * [placeholder for DX summary]
2061 ** user code with &REST lists declared dynamic-extent, under high
2062 speed or space and low safety and debug optimization policy.
2063 * The manual has been converted to Texinfo format and the debugger
2064 chapter from the cmucl manual has been added.
2065 * A facility has been added to extract documentation strings from
2066 sbcl and store them as Texinfo-formatted snippets for inclusion in
2067 the manual (via Texinfo's @include directive)
2068 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2069 the test case to Dave Roberts)
2070 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2071 pointers. (reported by Sean Ross)
2072 * bug fix: PROFILE output is printed nicely even for large numerical
2073 values. (thanks to Zach Beane)
2074 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2075 greater than 32 handle EOF correctly.
2076 * bug fix: on X86 an immediate argument of the IMUL instruction is
2077 correctly printed in disassembly. (thanks to Lutz Euler)
2078 * bug fix: class slots in redefined classes preserve their old
2079 values. (thanks to Bruno Haible and Nikodemus Siivola)
2080 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2081 succeeds. (reported by Marco Baringer)
2082 * fixed some bugs revealed by Paul Dietz' test suite:
2083 ** READ-SEQUENCE now works on ECHO-STREAMs.
2084 ** RATIONALIZE works more according to its specification. (thanks
2087 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2088 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2089 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2090 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2091 should matter to you only if you rebind the printer control
2092 variables and then find you want different bindings in the
2093 debugger than in the ordinary execution of your program.)
2094 * The runtime build system has been tweaked to support building
2095 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2096 assembler and linker. (thanks to Nikodemus Siivola)
2097 * Unbound, undefined, undeclared variables now trigger full
2098 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2099 more often programmer error than deliberate exploitation of undefined
2101 * optimization: the hash algorithm for strings has changed to one
2102 that is less vulnerable to spurious collisions. (thanks to Juho
2104 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2105 needless bounds checking. (thanks to Juho Snellman)
2106 * optimization: implemented multiplication as a modular
2107 (UNSIGNED-BYTE 32) operation on the PPC backend.
2108 * fixed some bugs revealed by Paul Dietz' test suite:
2109 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2111 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2112 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2113 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2115 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2117 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2119 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2120 their output stream on EOF from read.
2121 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2122 have been read to end-of-file.
2123 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2125 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2126 description of determination of which consecutive characters
2128 ** Printing the "Space" character with escaping on now yields "#\\ ",
2129 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2130 ** Reading floating-point numbers with *READ-BASE* set to a number
2131 less than 10 works correctly.
2132 ** Reading floating-point numbers with *READ-BASE* set to a number
2133 more than 10 works correctly.
2134 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2135 the readtable currently in effect.
2137 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2138 * minor incompatible change: parsing of namestrings on a physical
2139 (Unix) host has changed; numbers after the final #\. in a
2140 namestring are no longer interpreted as a version field. This is
2141 intented to be largely invisible to the user, except that the
2142 meaning of the namestring "*.*.*" has changed: it now refers to a
2143 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2144 should usually be replaced by
2145 (make-pathname :name :wild :type :wild :version :wild)
2146 with the added benefit that this is more likely to be portable.
2147 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2148 option to OPEN now signals an error if the file being opened
2149 exists; this may have an impact on existing code.
2150 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2151 signals received, on the PowerPC platforms (both Linux and
2152 Darwin). (thanks to Pierre Mai for pointing out the location of
2154 * several fixes on OS X: The system now builds and runs cleanly on
2155 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2156 (thanks to Brian Mastenbrook)
2157 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2158 second-resolution: integer multiples of 1/3600 between -24 and 24.
2159 (thanks to Vincent Arkesteijn)
2160 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2161 type when called with 1 argument; PEEK-CHAR checked type of
2162 PEEK-TYPE only after having read first character from a
2163 stream. (reported by Peter Graves)
2164 * bug fix: the garbage collector now has much better locality
2165 behaviour, and in particular no longer treats all memory as being
2166 exhausted when many small objects point to each other in a deeply
2168 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2169 recognized as being TYPEP their class.
2170 * bug fix: the PUSHNEW documentation string has been corrected.
2171 (thanks to Vincent Arkesteijn)
2172 * bug fix: defaulting of the value for the last of an atypically
2173 large number of multiple values being bound was not being
2174 performed correctly on the Alpha or PPC platforms
2175 * optimization: implemented multiplication as a modular
2176 (UNSIGNED-BYTE 32) operation on the x86 backend.
2177 * optimization: SEARCH on simple-base-strings can now be open-coded.
2178 (see also contrib/compiler-extras.lisp for inspiration for
2179 teaching the compiler about the Boyer-Moore algorithm).
2180 * value, returned by MAX (and MIN) called with several EQUALP, but
2181 not EQL, arguments now does not depend on compiler settings.
2182 * fixed some bugs revealed by Paul Dietz' test suite:
2183 ** in stack analysis liveness information is propagated from
2184 non-local entry points.
2185 ** pathwise CAST removing failed when the CAST node did not start
2187 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2188 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2190 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2191 host is already defined.
2192 ** RENAME-FILE works on streams instead of signalling an internal
2194 ** PEEK-CHAR uses the current readtable when determining whether
2195 or not a character is whitespace.
2196 ** MERGE-PATHNAMES handles the case when the pathname does not
2197 specify a name while the default-pathname specifies a version.
2198 ** Pathnames now stand a better chance of respecting print/read
2200 ** Attempting to use standardized file system operators with a
2201 pathname with invalid :DIRECTORY components signals a
2203 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2205 ** DIRECTORY on logical pathnames is more correct.
2206 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2207 signal a TYPE-ERROR if their argument is not a stream.
2208 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2209 designator argument does not designate a stream.
2210 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2211 examining the synonym.
2212 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2214 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2215 element-type larger than ([UN]SIGNED-BYTE 32).
2217 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2218 * When built with the :SB-FUTEX feature, threaded builds now take
2219 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2220 for faster/more reliable mutex and condition variable support.
2221 * Incompatible change (but one you probably shouldn't have been using
2222 anyway): the interface and code for arbitrating between multiple
2223 threads in the same user session has been redesigned.
2224 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2225 error when called without an explicit environment argument.
2226 (thanks to Nikodemus Siivola)
2227 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2228 confused about their position. (thanks to Adam Warner and Gerd
2230 * bug fix: Pretty printing backquoted forms with unquotations in the
2231 argument list position of various code constructs such as LAMBDA
2232 now works correctly. (reported by Paul Dietz)
2233 * bug fix: Pretty printing unquotations no longer loses all
2234 stream position information.
2235 * optimization: performance of string output streams is now less
2236 poor for multiple small sequence writes.
2237 * optimization: performance of CSUBTYPEP in the presence of complex
2238 expressions involving CONS and NOT many times has been improved.
2239 (reported by Paul Dietz)
2240 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2242 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2243 streams for the REPL, for the debugger, and so on. (thanks to
2245 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2246 * fixed some bugs revealed by Paul Dietz' test suite:
2247 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2248 is ignored if it is too small, rather than propagating through
2249 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2250 ** extremely complex negations of CONS types were not being
2251 sufficiently canonicalized, leading to inconsistencies in
2253 ** VALUES tranformer lost derived type.
2255 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2256 * fixed a bootstrapping bug: the build process no longer assumes
2257 that the various BOOLE-related constants have the same value in
2258 host and target lisps. (noted by Paul Dietz' test suite on an
2259 SBCL binary built from CLISP)
2260 * The system can now be dynamically linked on the MIPS platform,
2261 which enables dynamic loading of foreign code from Lisp. (thanks
2262 to Ralf Baechle for discussions on the MIPS ABI)
2263 * The system now records debugging information for its own source
2264 files in a filesystem-position-independent manner, relative to
2265 the "SYS" logical host.
2266 * fixed a compiler bug: MV-LET convertion did not check references
2267 to the "max args" entry point. (reported by Brian Downing)
2268 * tweaked disassembly notes to be less confident about proclaiming
2269 some instruction as an LRA. (thanks to Brian Downing)
2270 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2271 now each have their own history, command character, and other
2272 characteristics. (thanks to David Lichteblau)
2273 * fixed some bugs revealed by Paul Dietz' test suite:
2274 ** compiler failure in compiling LOGAND expressions including a
2276 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2277 shift greater than 32.
2278 ** FLUSH-DEST did not mark blocks for type check regeneration.
2279 ** HANDLER-CASE failed to accept declarations in handler clauses
2280 in some circumstances.
2282 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2283 * New code in contrib/sb-introspect (still probably not entirely
2284 stable yet) provides some support for smart Lisp development
2285 environments like SLIME.
2286 * The conditions signalled for errors occurring when loading .fasl
2287 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2288 in a way which should help ASDF recover gracefully.
2289 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2290 stuff has been cleaned up. If you code contrib/ stuff, this might
2291 affect you, and you can look at contrib/README, contrib/STANDARDS,
2292 and/or the 0.8.4.27 diff to check.
2293 * In full calls the compiler now does not generate checks for declared
2294 argument types for all arguments.
2295 * various threading fixes
2296 ** and some experimental patches which didn't make it into
2297 the main tree for this release, but which are shipped in
2298 contrib/experimental-thread.patch as a possible fix for some
2299 failures (deadlock, spinning...) in GC-intensive multithreaded
2301 * fixed PPC build problem (source code incompatibility of different
2302 library versions): added offsetof() hackery which attempts to divine
2303 where glibc maintainers put uc_mcontext today
2304 * fixed bug 282: compiler does not trust type assertions while passing
2305 arguments to a full call.
2306 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2307 &OPTIONAL VALUES type parameter.
2308 * fix bug 214: algorithm for noting rejected templates is now more
2309 similar to that of template seletion. (also reported by rydis on
2311 * fixed bug 141b: printing backquoted information readably and prettily
2312 inserts a space where necessary.
2313 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2314 (CHARACTER 10) are now reported as errors, rather than propagated
2315 as unknown types. (reported by piso on #lisp)
2316 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2317 with values NIL and :ERROR. (thanks to Milan Zamazal)
2318 * fixed bug 191c: CLOS now does proper keyword argument checking as
2319 described in CLHS 7.6.5 and 7.6.5.1.
2320 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2321 longer raise an error; further, using a list as a for-as-arithmetic
2322 counter now raises a meaningful error.
2323 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2324 now understood by sequence creation functions such as MAKE-SEQUENCE
2326 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2327 a STRING-INPUT-STREAM.
2328 * compiler enhancement: SIGNUM is now better able to derive the type
2330 * type declarations inside WITH-SLOTS are checked. (reported by
2332 * fixed some bugs revealed by Paul Dietz' test suite:
2333 ** incorrect optimization of TRUNCATE for a positive first
2334 argument and negative second.
2335 ** compiler failure in let-convertion during flushing dead code.
2336 ** compiler failure while deriving type of TRUNCATE on an
2337 interval, containing 0.
2338 ** ASH of a negative bignum by a negative bignum count now returns
2340 ** intersection of CONS types now canonicalizes properly, fixing
2341 inconsistencies in SUBTYPEP.
2343 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2344 * incompatible change: The --disable-debugger command line
2345 option now clobbers the debugger at a more fundamental
2346 level, by redefining #'INVOKE-DEBUGGER instead of by
2347 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2348 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2349 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2350 enter the debugger REPL and then suffer recursive errors
2351 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2352 while under the new system, BREAK in --disable-debugger mode
2353 terminates the system just as an unhandled error would.
2354 * fixed compiler performance when processing loops with a step >1;
2355 * bug fix: DOCUMENTATION now retrieves generic function
2356 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2357 support has been systematized, and now supports the methods
2358 specified by ANSI, along with a default method and a method for
2359 slot documentation. (reported by Nathan Froyd)
2360 * bug fix: effective methods associated with a generic function are
2361 no longer cached over a change of that generic function's method
2362 combination. (reported by Andreas Fuchs)
2363 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2364 names a non-existent directory. (thanks to Andreas Fuchs)
2365 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2366 the PPC platform, lead to stack corruption; this has been fixed.
2367 (reported by Rainer Joswig)
2368 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2369 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2370 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2371 arithmetic in general, is now much more reliable on the PPC
2373 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2374 platform now returns the right answer.
2375 * optimization: restored some effective method precomputation in
2376 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2377 precomputation is now tunable.
2378 * optimization: compiler-internal data structure use has been
2379 reviewed, and changes have been made that should improve the
2380 performance of the compiler by about 20%.
2381 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2382 simple-base-strings and simple-bit-vectors is improved.
2383 * optimization: the optimization of 32-bit logical and arithmetic
2384 functions introduced in version 0.8.3 on the x86 has been
2385 implemented on the mips, ppc and sparc platforms; an
2386 implementation of the same facility, but for 64-bit arithmetic,
2387 has been added for the alpha.
2388 * microoptimization: the compiler is better able to make use of the
2389 x86 LEA instruction for multiplication by constants.
2390 * bug fix: in some situations compiler did not report usage of
2391 generic arithmetic in (SPEED 3) policy.
2392 * bug 145b fix: compiler used wrong type specifier while converting
2393 MEMBER-types to numeric.
2394 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2396 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2397 index argument now works properly on the Alpha platform.
2398 * bug fix: floating point exception treatment on the Alpha platform
2400 * bug fix: FILE-POSITION works much better on string input and
2401 output streams. (thanks to Nikodemus Siivola)
2402 * bug fix: many threading/garbage collection symptoms sorted.
2403 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2404 might be pseudo-atomic.
2405 * internal change: Stopping for GC is now done with signals not
2406 ptrace. GC is now done in whichever thread wanted it, instead of
2408 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2410 * bug fix: result form in DO is not contained in the implicit
2412 * incompatible change: ICR structure is changed; the value part of
2413 CONTINUATION is now called LVAR; corresponding functions are
2414 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2415 * added type deriver for ISQRT (thanks to Robert E. Brown).
2416 * bug fix: better support for loading from the command line when an
2417 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2419 * fixed some bugs revealed by Paul Dietz' test suite:
2420 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2421 ** ROUND and FROUND now give the right answer when given very
2422 small float arguments.
2423 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2425 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2426 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2427 ** LCM should return a non-negative integer.
2428 ** PARSE-INTEGER returned the index of a terminator instead of the
2429 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2430 ** PARSE-INTEGER returned an incorrect index being applied to a
2432 ** LCM with two arguments of 0 returns 0 rather than signalling
2434 ** unsigned addition of a 32-bit constant with the high bit set no
2435 longer causes an internal compiler error on the x86.
2436 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2437 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2438 with negative last argument.
2439 ** byte specifiers with zero size and position no longer cause
2440 an error during type derivation.
2441 ** bignum multiplication on the Alpha platform now returns the
2443 * porting: The system now builds on SuSE AMD64, although it still
2444 generates a 32-bit binary.
2445 * .fasl file incompatibility: The fasl file version number has
2446 been incremented (because of the changes to internal compiler
2447 data structures referred to above).
2449 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2450 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2451 more accurately, on the Darwin kernel running on PowerPC hardware.
2452 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2453 * Compiler code deletion notes now signal a condition of type
2454 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2455 an associated MUFFLE-WARNING restart.
2456 * The compiler now performs limited argument count validation of
2457 constant format strings in FORMAT, and where appropriate in ERROR,
2458 CERROR and WARN. (thanks to Gerd Moellmann)
2459 * New ASDF-INSTALL contrib can be used for automatic download and
2460 installation of third-party Lisp code from CCLAN or other sites
2462 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2463 which forces another thread to execute a function supplied by the
2465 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2466 now accept and act upon their :ELEMENT-TYPE keyword argument.
2467 (reported by Martin Atzmueller, Edi Weitz)
2468 * bug fix: FILE-POSITION now accepts position designators up to
2469 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2470 is the greater. (thanks to Patrik Nordebo)
2471 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2472 Kalle Olavi Niemitalo)
2473 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2474 printable. (reported by Eric Marsden)
2475 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2476 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2477 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2478 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2479 (thanks to Stig Sandoe)
2480 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2481 UNREAD-CHAR now correctly includes the unread character in the
2482 target sequence. (thanks to Gerd Moellmann)
2483 * bug fix in threaded builds: the system can now be suspended and
2484 resumed by shell job control with minimal disruption.
2485 * bug fixes in times and timezones >2038AD
2486 * better handling of "where is GNU make?" problem in build scripts
2487 (thanks to Nikodemus Siivola)
2488 * new optimization: inside a named function any reference to a
2489 function with the same name is considered to be a self-reference;
2490 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2491 optimization quality.
2492 * new optimization on x86: logical functions and + now have
2493 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2494 used when the result is truncated to 32 bits.
2495 * VALUES declaration is partially enabled.
2496 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2497 * fixed some bugs revealed by Paul Dietz' test suite:
2498 ** The system now obeys the constraint imposed by
2499 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2500 types form a lattice under type intersection.
2501 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2502 ** ASSOC now ignores NIL elements in an alist.
2503 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2504 and (1+ MOST-POSITIVE-FIXNUM) answers.
2505 ** The addition of a method with invalid qualifiers to a generic
2506 function does not cause an error to be signalled immediately;
2507 a warning is signalled, and the error is generated only on
2508 calling the generic function.
2509 * changed .fasl file version number, in part to add type codes for
2510 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2511 obscure ANSI requirements
2513 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2514 * fixed bug 148: failure to inline-expand a local function left
2515 garbage, confusing the compiler.
2516 * fixed bugs 3cd: structure slot readers perform type check if the
2517 slot can have an invalid value (i.e. it is either not initialized
2518 or can be written with a less specific slot writer).
2519 * bug fix: the compiler now traps array references to elements off
2520 the end of an array; previously, the bounds checking in some
2521 circumstances could go off-by-one.
2522 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2524 * type declarations for array element types now obey the description
2525 on the CLHS page "Declaration TYPE", as per discussions on
2526 sbcl-help around 2003-05-08. This means that a declaration
2527 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2528 declaration, all references to BAR will be asserted or assumed
2529 (with THE, so dependent on compiler policy) to involve objects of
2530 type FOO. Note that no such declaration is implied in
2531 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2532 * declared types of functions from the "Conditions"
2533 chapter. (reported by Paul Dietz)
2534 * bug fix: CERROR accepts a function as its first argument.
2535 * bug fix: NTH an NTHCDR accept a bignum as index
2536 arguments. (reported by Adam Warner)
2537 * optimization: character compare routines now optimize comparing
2538 against a constant character. (reported by Gilbert Baumann)
2539 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2540 argument now works properly on the MIPS platform.
2541 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2542 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2543 declared to have a functional type, does not cause an error
2545 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2546 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2547 * fixed bug: if last continuation of a deleted block has a
2548 destination, this destination should be deleted too. (reported by
2550 * fixed a bug in the bootstrap process: the host compiler's values
2551 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2552 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2553 test case from Patrik Nordebo)
2554 * improved the ability of the disassembler on the PPC platform to
2555 provide helpful disassembly notes.
2556 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2557 the class in more cases than previously.
2558 * bug fix: FILE-POSITION now understands :START and :END for
2559 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2560 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2561 8) data. (thanks to David Lichteblau)
2562 * bug fix: it is possible to add a method to a generic function
2563 without lambda list.
2564 * bug fix: reader failed to signal END-OF-FILE inside an
2565 object representation. (reported by Nikodemus Siivola)
2566 * fixed some bugs revealed by Paul Dietz' test suite:
2567 ** LAST and [N]BUTLAST should accept a bignum.
2568 ** condition slot accessors are methods.
2569 ** (VECTOR NIL) is a subtype of STRING.
2571 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2572 * minor incompatible change: some nonsensical specialized lambda
2573 lists (used in DEFMETHOD) which were previously ignored now signal
2575 * minor incompatible change: the system is now aware of the types of
2576 variables in the COMMON-LISP package, and will signal errors for
2577 most violations of these type constraints (where previously they
2578 were silently accepted).
2579 * minor incompatible change: COMPILE-FILE now uses the freedom
2580 afforded (ANSI 3.2.2.3) to use derived function types for
2581 functions defined in the same file. This also permits the system
2582 to warn on static type mismatches and function
2583 redefinition. (Currently it does not work with high DEBUG level.)
2584 * minor incompatible change: VALUES declaration is disabled.
2585 * When issuing notes, the compiler now signals a condition of type
2586 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2587 restart for use in user handlers. It is expected that the
2588 COMPILER-NOTE condition will eventually become a condition
2589 supertype to a hierarchy of note types, which will then be
2590 handleable in a similar fashion. However, at the moment, no such
2591 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2592 but it's an implementation detail, not a classification for the
2594 * Changes in type checking closed the following bugs:
2595 ** type checking of unused values (192b, 194d, 203);
2596 ** template selection based on unsafe type assertions (192c, 236);
2597 ** type checking in branches (194bc).
2598 * A short form of VALUES type specifier has ANSI meaning (it has
2599 increased the number of situations when SBCL cannot perform type
2601 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2602 keywords or constants is permissible.
2603 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2604 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2605 classes. (thanks to Antonio Martinez)
2606 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2607 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2608 argument to a pretty-print stream. Instead, it leaves any such
2609 operations to DESCRIBE-OBJECT methods.
2610 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2612 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2613 lambda lists. (thanks to David Lichteblau)
2614 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2615 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2616 (thanks to Gerd Moellmann)
2617 * fixed bug 246: increased compilation speed of long
2618 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2620 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2622 * DEFCONSTANT now throws a condition of type
2623 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2624 constant to a non-EQL value; CONTINUE and ABORT restarts
2625 respectively change and preserve the value.
2626 * fixed bug 63: The code walker, part of the implementation of CLOS,
2627 is now better at handling symbol macros.
2628 * bug fix: There is no longer an internal implementation type named
2629 CL:LENGTH. (reported by Raymond Toy)
2630 * bug fix: In macro-like defining macros/special operators the
2631 implicit block does not enclose the lambda list.
2632 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2633 enclosed in parentheses) are not suitable as type specifiers, and
2634 their use properly signals an error now.
2635 * bug fix: An argument count mismatch for a type specifier in code
2636 being compiled no longer causes an unhandled error at compile
2637 time, but signals a compile-time warning.
2638 * fixed simple vector readable printing
2639 * bug fix: DESCRIBE takes more care over whether the class
2640 precedence list slot of a class is bound before accessing it.
2641 (reported by Markus Krummenacker)
2642 * bug fix: FORMATTER can successfully compile pretty-printer format
2643 strings which use variants of the ~* directive inside.
2644 * bug fix: SEARCH now applies its TEST predicate to the elements of
2645 the arguments in the correct order. (thanks to Wolfhard Buss)
2646 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2647 in type inference. (thanks to Robert E. Brown)
2648 * bug fix: Reading in symbols with an explicit package name of ""
2649 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2650 * fixed some bugs revealed by Paul Dietz' test suite:
2651 ** NIL is now allowed as a structure slot name.
2652 ** Arbitrary numbers, not just REALs, are allowed in certain
2653 circumstances in LOOP for-as-arithmetic clauses.
2654 ** Multiple class redefinitions before slot access no longer
2655 causes a type error.
2656 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2657 association between the name and a class.
2658 ** Generic functions with non-standard method-combination and over
2659 five methods all of which return constants no longer return NIL
2660 after the first few invocations. (thanks to Gerd Moellmann)
2661 ** CALL-NEXT-METHOD with no arguments now passes the original
2662 values of the arguments, even in the presence of assignment.
2663 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2665 ** DISASSEMBLE works with closures and funcallable instances.
2666 ** ADD-METHOD now returns the generic function, not the new method.
2667 ** FIND-METHOD signals an error if the lengths of the specializers
2668 is incompatible with the generic function, even if the ERRORP
2670 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2671 which its argument is a member.
2672 ** DEFCLASS only redefines the class named by its class-name
2673 argument if that name is the proper name of the class;
2674 otherwise, it creates a new class.
2675 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2676 of the UNBOUND-SLOT condition to the name of the slot.
2677 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2678 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2679 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2680 SLOT-MAKUNBOUND in the specified fashion.
2682 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2683 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2684 cross-compilation host. As a consequence, we can now bootstrap our
2685 way up to SBCL starting with a bare gcc toolchain and human-readable
2686 source code (first the source to CLISP, then the source to SBCL).
2687 * A contributed module containing a partial implementation of the
2688 simple-streams interface has been included. (thanks to Rudi
2690 * A contributed module implementing the RFC1321 Message Digest
2691 Algorithm, known as MD5, has been included.
2692 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2693 no longer has any effect, as the code controlled by this feature
2694 has been deleted. (As far as we know, no-one has ever built using
2695 this feature, and its semantics were confused in any case).
2696 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2697 work on conditions (as required by the ANSI specification),
2698 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2699 expected behaviour on conditions. Users should note, however,
2700 that such behaviour is not required by the ANSI specification,
2701 and so use of this behaviour may render their code unportable.
2702 * fixed some bugs revealed by Paul Dietz' test suite:
2703 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2705 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2706 ** SXHASH on condition objects no longer returns NIL.
2707 ** :ALLOCATION :CLASS slots are better treated; their values are
2708 updated on class redefinition, and initforms inherited from
2709 superclasses are applied.
2710 ** REMOVE-METHOD returns its generic function argument even when
2711 no method was removed.
2712 ** SHARED-INITIALIZE now initializes the values of the requested
2713 slots, including those with :ALLOCATION :CLASS.
2714 ** ALLOCATE-INSTANCE now works on structure classes defined via
2715 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2717 ** SLOT-EXISTS-P now works on conditions, as well as structures
2719 ** MAKE-LOAD-FORM now has the required methods on
2720 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2721 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2722 keyword, and now implements the SLOT-NAMES argument.
2723 ** methods with &OPTIONAL arguments no longer allow too many
2724 arguments to be passed in the call without error.
2725 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2726 option is consistent with the required arguments of the generic
2727 function lambda list.
2728 * bug fix: REQUIRE accepts a string designator. (Thanks to
2730 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2731 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2732 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2733 * bug fix: The compiler now checks for duplicated variables in macro
2735 * bug fix: SETQ on globals returns the correct value.
2736 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2737 not a COMPILER-ERROR (followed by some other strange error on
2738 choosing the CONTINUE restart).
2739 * bug fix: make.sh and friends are now more consistent in the way that
2740 they look for GNU "make".
2742 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2743 * experimental native threads support (on x86 Linux >=2.4 only).
2744 This is not compiled in by default: you need to add :SB-THREAD to
2745 the target features. See the "Beyond ANSI" chapter of the manual
2747 * fix for longstanding nonANSIism: The old distinction between
2748 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2749 The return value from CL:FIND-CLASS is now a CLOS class, and
2750 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2751 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2752 * An interface to the MetaObject Protocol, as described in Kiczales,
2753 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2754 MIT Press, 1991, is available from the SB-MOP package.
2755 * incompatible change: the SB-PCL package should now be considered
2756 a private implementation detail, and no longer a semi-private MOP
2758 * minor incompatible change: due to rearrangement for threads, the
2759 control stack and binding stack are now allocated at arbitrary
2760 addresses instead of being hardcoded per-port. Users affected by
2761 this probably have to be doing advanced things with shared
2762 libraries, and will know who they are.
2763 * minor incompatible change: Previously, all --eval forms used were
2764 processed with READ before any of them were processed with EVAL.
2765 Now each --eval form is processed with both READ and EVAL before
2766 the next --eval form is processed. (Thus package operations like
2767 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2768 work as the user might reasonably expect.)
2769 * minor incompatible change: *STANDARD-INPUT* is now only an
2770 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2772 * minor incompatible change: Y-OR-N-P is now character-oriented,
2773 not line oriented. Also, YES-OR-NO-P now works without errors.
2774 (thanks to Antonio Martinez)
2775 * sb-aclrepl module improvements: an integrated inspector, added
2776 repl features, and a bug fix to :trace command.
2777 * Known functions, which cannot be open coded by the backend, are
2778 considered to be able to check types of their arguments. (fixing
2779 a bug report by Nathan J. Froyd)
2780 * fixed a bug in computing method discriminating functions: It is
2781 now possible to define methods specialized on classes which have
2782 forward-referenced superclasses. (thanks to Gerd Moellmann)
2783 * fixed evaluation order in optional entries (reported by Gilbert
2785 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2786 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2787 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2788 required-by-AMOP NAME argument, as well as CLASS and
2789 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2790 * fixed bug 20: DEFMETHOD can define methods using names that are
2791 not the proper names of classes to designate class specializers.
2792 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2793 Unix file descriptors, instead of blowing up. (thanks to Antonio
2795 * Garbage collection refactoring: user-visible change is that a
2796 call to the GC function during WITHOUT-GCING will not do garbage
2797 collection until the end of the WITHOUT-GCING. If you were doing
2798 this you were probably losing anyway.
2799 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2800 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2801 TYPEP the latter but not the former.
2802 * The compiler issues a full WARNING for calls to undefined functions
2803 with names from the CL package.
2804 * MAP-INTO for a vector destination is open coded. (reported by
2805 Brian Downing on c.l.l)
2806 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2807 documentation string.
2808 * fixed some bugs revealed by Paul Dietz' test suite:
2809 ** COPY-ALIST now signals an error if its argument is a dotted
2811 ** Condition slots are now accessed more correctly in the presence
2812 of multiple initargs for a given slot.
2813 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2814 exclude restarts of the same name associated with a different
2816 ** DEFCLASS of forward-referenced classes with another
2817 forward-referenced class in the superclasses list no longer
2819 ** Condition slots are now initialized once each, not multiple
2820 times. (thanks to Gerd Moellmann)
2821 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2823 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2824 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2825 arguments contain duplicated elements.
2826 ** RESTART-CASE understands local macros.
2827 ** RESTART-CASE associates exactly its own restarts with a condition.
2828 ** ENDP in safe mode checks its argument to be of type LIST.
2829 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2830 in question is unbound.
2831 ** Optimized MAKE-INSTANCE functions no longer cause internal
2832 assertion failures in the presence of duplicate initargs.
2833 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2834 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2835 propagated even in the ctor optimized implementation of
2837 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2839 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2840 * a better implementation of SXHASH on (simple) bit vectors,
2841 measured both in execution speed and in distribution of results
2842 over the positive fixnums, has been installed. Likewise, a better
2843 implementation of EQUAL for simple bit vectors is now available.
2844 * fixed CEILING optimization for a divisor of form 2^k.
2845 * fixed bug 240 (emitting extra style warnings "using the lexical
2846 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2847 by Antonio Martinez)
2848 * fixed SXHASH, giving different results for NIL depending on type
2849 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2850 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2851 variable. (found by Rolf Wester)
2852 * SBCL does not ignore type declarations for special
2853 variables. (reported by rif on c.l.l 2003-03-05)
2854 * some bug fixes in contrib/sb-aclrepl/
2855 * fixed some bugs revealed by Paul Dietz' test suite:
2856 ** a bug in the CONS type specifier, whereby the CAR and CDR
2857 types got intertwined, has been fixed;
2858 ** the type system is now able to reason about the interaction
2859 between INTEGER and RATIO types more completely;
2860 ** APPEND, [N]REVERSE and NRECONC check that those their
2861 arguments, which must be proper lists, are really so;
2862 ** An array specialized to be unable to hold elements has been
2863 implemented, as required -- yes, really -- by ANSI;
2864 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2865 on malformed property lists;
2867 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2868 * incompatible packaging change: in line with Unix convention,
2869 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2870 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2871 * REQUIRE and PROVIDE are now optionally capable of doing something
2872 useful. See the documentation string for REQUIRE.
2873 * infrastructure for a managed SBCL contrib system: contributed
2874 modules in this release include:
2875 ** the ASDF system definition facility;
2876 ** an interface to the BSD Sockets API;
2877 ** an ACL-like convenience interface to the repl;
2878 (thanks to Kevin Rosenberg)
2879 ** an implementation of ROTATE-BYTE, with efficient implementation
2881 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2882 gives a closer approximation to the right answer than previously.
2883 (thanks to Raymond Toy)
2884 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2885 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2886 optional environment arguments, as required by ANSI.
2887 * fixed bugs in other functions taking environment objects, allowing
2888 calls with an explicit NIL environment argument to be compiled
2890 * fixed bug 228: primary return values from
2891 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2892 COMPILE or FUNCTION.
2893 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2894 :TYPE will no longer signal errors on innocuous objects.
2895 * fixed bug 231b: SETQ is better at respecting type declarations in
2896 the lexical environment.
2897 * fixed a bug in DEFCLASS: classes named by symbols with no or
2898 unprintable packages can now be defined.
2899 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2900 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2901 * fixed compiler failure related to checking types of functions.
2902 (reported by Robert E. Brown)
2903 * the compiler is now much more consistent in its error-checking
2904 treatment of bounding index arguments to sequence functions: in
2905 (SAFETY 3) code, errors will be signalled in almost all cases if
2906 invalid sequence bounding indices are passed to functions defined
2907 by ANSI to operate on sequences.
2908 * fixed a bug in the build procedure: documentation of SBCL-specific
2909 packages is now preserved and available in the final Lisp image.
2910 * lifted FDEFINITION lookup out of loops in the implementation of
2911 many list operations. (thanks to Robert E. Brown)
2912 * fixed a bug in the reader: the #n# reader macro now works for
2913 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2914 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2915 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2916 E. Brown and Tony Martinez respectively)
2917 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2918 SLOT-BOUNDP: in particular, it is now able to optimize them much
2919 better, and is now not vulnerable to having packages renamed.
2920 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2921 always called when a slot is not present in an instance. (thanks
2923 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2924 longer signal an error on unreading a character following EOF on
2925 the previous constituent stream. (thanks to Tony Martinez)
2926 * fixed some bugs revealed by Paul Dietz' test suite:
2927 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2928 not just nonnegative fixnums;
2929 ** the logical bit-array operators such as BIT-AND now accept an
2930 explicit NIL for their "opt-arg" argument (to indicate a
2931 freshly-consed result bit-array);
2932 ** ELT now signals an error on an invalid sequence index in safe
2934 ** the type system is now cleverer about negations of numeric
2935 types, and consequently understands the BIGNUM and RATIO types
2937 ** the type system is now cleverer about the interaction between
2938 INTEGER and RATIO types: while bugs still remain, many more
2939 cases are accurately computed;
2940 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2941 if it is in the last clause;
2942 ** CONSTANTLY now correctly returns a side-effect-free function in
2944 ** DECLARE is no longer treated as a special-operator; in
2945 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2946 * incremented fasl file version number due to the change in the
2947 DEFSTRUCT-SLOT-DESCRIPTION structure.
2949 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2950 * minor incompatible change: code processed by the "interpreter" or
2951 EVAL now has a compilation optimization policy of (DEBUG 2)
2952 (changed from (DEBUG 1)) to improve debuggability of interactive
2953 development, and to allow the use of the debug RETURN command in
2955 * an experimental implementation of the RETURN command for the
2956 debugger has been included. (thanks to Frederik Kuivinen)
2957 * fixed bug 62: constraints were not propagated into a loop.
2958 * fixed bug in embedded calls of SORT (reported and investigated by
2960 * fixed some bugs revealed by Paul Dietz' test suite:
2961 ** printing and reading of arrays with some dimensions having
2962 length 0 (thanks to Gerd Moellmann);
2963 ** BOA constructor with &AUX argument without a default value does
2964 not cause a type error;
2965 ** CONSTANTP now returns true for all self-evaluating objects.
2967 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2968 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2969 accessors that are related by inheritance, as specified in the
2970 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2971 Valtteri Vuorikoski)
2972 * The compiler is now able to inline functions that were defined in
2973 a complex lexical environment (e.g. inside a MACROLET).
2974 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2976 * Support for the upcoming FreeBSD-5.0 release has been included.
2977 (thanks to Dag-Erling Smorgrav)
2978 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2979 effect when it is not in a toplevel context.
2980 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2981 stand a better chance of being correct. (thanks to Gerd
2983 * fixed bug in COERCE, which now signals an error on coercing a
2984 rational to a bounded real type which excludes the expected
2986 * The compiler is now able to derive types more accurately from the
2987 COERCE and COMPILE functions.
2988 * fixed bug 223: functional binding is considered to be constant
2989 only for symbols in the CL package.
2990 * fixed bug 231: SETQ did not check the type of a variable being set
2991 (reported by Robert E. Brown)
2992 * A new optimization for MAKE-INSTANCE has been included, fixing
2993 various bugs (including relating to :ALLOCATION :CLASS slots and
2994 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2996 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2997 ** As required by ANSI, LOOP now disallows anonymous collection
2998 clauses such as COLLECT I in conjunction with aggregate boolean
2999 clauses such as THEREIS (= I 1);
3000 ** LOOP now signals an error when any variable is reused in the
3001 same loop (including the potentially useful construct analogous
3002 to WITH A = 1 WITH A = (1+ A);
3003 ** IT is only a special loop symbol within the first clause of a
3004 conditional loop clause;
3005 ** LOOP with a typed iteration variable over a hashtable now
3006 signals a type error iff it should.
3007 * fixed some other bugs revealed by Paul Dietz' test suite:
3008 ** FILE-STREAM now names the class previously known as FD-STREAM;
3009 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3010 argument) no longer signals an error;
3011 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3012 of the slot symbol, rather than using the current package
3013 ((:CONC-NAME "") continues to intern the slot's name in the
3015 * incremented fasl file version number, because of the incompatible
3016 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3017 of the new implementation of DEFINE-COMPILER-MACRO.
3019 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3020 * Support for building SBCL for MIPS platforms running in
3021 little-endian mode has now been checked in, and basic
3022 functionality on said platforms verified.
3023 * minor incompatible change: PCL now records the pathname of a file
3024 in which methods and the like are defined, rather than its
3026 * minor incompatible change: TRUENAME now considers the truename of
3027 a file naming a directory to be the pathname with :DIRECTORY
3028 component indicating that directory.
3029 * minor incompatible change: a NAMED clause in the extended form of
3030 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3031 reason for the previous behaviour is unclear.
3032 * more systematization and improvement of CLOS and MOP conformance
3033 in PCL (thanks to Gerd Moellman and Pierre Mai):
3034 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3036 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3037 primary methods with no specializers;
3038 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3040 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3041 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3042 FORWARD-REFERENCED-CLASSes; error reporting on
3043 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3045 ** SXHASH on CLOS instances now uses a slot internal to the
3046 instance to return different numbers on distinct instances,
3047 while preserving the same return value through invocations of
3049 ** DEFMETHOD signals errors when methods with longer incongruent
3050 lambda lists are added to generic functions;
3051 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3052 CLASS, as specified in AMOP;
3053 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3054 on the order returned by the primary method for classes of
3055 class STANDARD-CLASS;
3056 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3057 * fixed some bugs shown by Paul Dietz' test suite:
3058 ** DOLIST puts its body in TAGBODY;
3059 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3061 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3062 value producing form;
3063 ** if more variables are given to PROGV than values, extra
3064 variables are bound and made to have no value;
3065 ** NSUBSTITUTE on list arguments gets the right answer with
3067 ** ELT signals an error of type TYPE-ERROR when the index argument
3068 is not a valid sequence index;
3069 ** LOOP signals (at macroexpansion time) an error of type
3070 PROGRAM-ERROR when duplicate variable names are found;
3071 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3072 ** FUNCALL of special-operators now cause an error of type
3074 ** PSETQ now works as required in the presence of side-effecting
3075 symbol-macro places;
3076 ** NCONC accepts any object as its last argument;
3077 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3079 ** loop-for-as-package does not require a package to be explicitely
3081 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3083 * fixed bug 166: compiler preserves "there is a way to go"
3084 invariant when deleting code.
3085 * fixed bug 172: macro lambda lists with required arguments after
3086 &REST arguments now cause an error to be signalled. (thanks to
3088 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3089 bug. (thanks to Matthew Danish)
3090 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3092 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3093 arguments are lexically rebound. (thanks to Gerd Moellmann and
3095 * fixed bug 194: error messages are now more informative when there
3096 is no primary method applicable in a call to a generic
3097 function. (thanks to Gerd Moellmann)
3098 * fixed bug in command line argument checking (thanks to Julian
3100 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3101 (thanks to Matthew Danish)
3102 * incremented fasl file version number, because of the
3103 SXHASH-related changes in the layout of CLOS data structures
3105 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3106 * minor incompatible change: The runtime (the Unix executable named
3107 "sbcl") is now much pickier about the .core files it will load.
3108 Essentially it now requires .core files to descend from the same
3109 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3110 as the runtime does. (The intent is to prevent the crashes which
3111 can occur, and which can even be reported as mysterious failures,
3112 when people patch the sources or change the build parameters
3113 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3114 sbcl and .core files.)
3115 * fixed bug: VALUES-LIST is no longer optimized away.
3116 * fixed bug 142: The FFI conversion of C string values to Lisp
3117 string values no longer conses excessively. (thanks to Nathan
3118 Froyd porting Raymond Toy's fix to CMU CL)
3119 * began to systematize and improve MOP conformance in PCL (thanks to
3120 Nathan Froyd, Gerd Moellman and Pierre Mai):
3121 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3123 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3124 ** FINALIZE-INHERITANCE is now called on class finalization;
3125 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3126 argument precedence order.
3127 * fixed bug 202: The compiler no longer fails on functions whose
3128 derived types contradict their declared type.
3129 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3130 so it can be non-toplevel.
3131 * The fasl file version number has changed (because of the new
3132 implementation of DEFMACRO).
3133 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3134 safe code, that any length requirement by their type-specifier
3135 argument is valid. The exceptions to this are described in bug
3137 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3138 functions now check the types of their inputs as required by ANSI.
3139 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3140 introduced symbol is DECLAREd to be SPECIAL.
3141 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3142 * fixed bug: :COUNT argument to sequence functions may be negative
3143 * fixed bug: body of DO-SYMBOLS may contain declarations
3144 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3145 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3146 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3147 symbol macro only once
3148 * fixed printing of call frame when argument list is unavailable
3149 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3150 * compiler no longer signals WARNING on unknown keyword
3153 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3154 * A beta-quality port to the mips architecture running Linux,
3155 based on the old CMUCL backend, has been made. It has been tested
3156 on a big-endian kernel, and works sufficiently well to be able to
3157 rebuild itself; it has not been tested in little-endian mode.
3158 * fixed an inconsistency between gencgc.c and purify.c which made
3159 dumping/loading .core files unreliable
3160 * fixed bug 120a: The compiler now deals correctly with IFs where
3161 the consequent is the same as the alternative, instead of
3162 misderiving the return type. (thanks to Alexey Dejneka)
3163 * fixed bug 113: Logical pathnames are now dumpable (the logical
3164 host is resolved at load-time, throwing an error if it is not
3166 * fixed bug 174: FORMAT's error message is slightly clearer when a
3167 non-printing character is used in a format directive.
3168 * fixed several bugs in compiler checking of type declarations, i.e.
3169 violations of the Python "declarations are assertions" principle
3170 (thanks to Alexey Dejneka)
3171 * fixed several bugs in PCL's error checking (thanks to Gerd
3173 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3175 * fixed bug in compilation of functions as first class values
3176 (thanks to Antonio Martinez-Shotton)
3177 * The compiler's handling TYPE-ERRORs which it can prove will
3178 inevitably happen at runtime has been cleaned up and corrected
3179 in several ways. (thanks to Alexey Dejneka)
3180 * improved argument type checking for various basic arithmetic
3181 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3182 lavished on them in the past that they can be compiled in many
3183 ways in different special cases
3184 * fixed bug 181: compiler checks validity of user supplied type
3186 * cleaned up code flushing in optimization: Function calls which
3187 should signal errors for safety purposes (e.g. which ANSI says
3188 should signal errors when their arguments are of incorrect type)
3189 are no longer optimized away.
3190 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3191 * incremented fasl file version number, because changes in the
3192 implementation of sequence functions like COERCE caused
3193 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3194 in old inline expansions) to become undefined. (Actually these
3195 changes were later undone, so we might very well be binary
3196 compatible with 0.7.7 after all, but leaving the version number
3197 incremented seemed like the simplest and most conservative
3200 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3201 * An alpha-quality port to the parisc architecture running Linux,
3202 based on the old CMUCL backend, has been made. This, even more so
3203 than the other backends, should be considered still a work in
3204 progress; known problems include that the Linux kernel in 64-bit
3205 mode does not propagate the correct sigcontext structure to
3206 userspace, and consequently SBCL on a parisc64 kernel will not
3208 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3209 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3210 suppressed.) Also now that the compiler is looking at declarations
3211 in the environment, it checks optimization declarations as well,
3212 and suppresses inlining when (> DEBUG SPEED).
3213 * More fixes have been made to treatment of floating point exception
3214 treatment and other Unix signals. In particular, floating point
3215 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3216 * The detection and handling of control stack exhaustion (infinite
3217 or very deeply nested recursion) has changed. Stack exhaustion
3218 detection is now done by write-protecting pages at the OS level
3219 and applies at all optimization settings; when found, a
3220 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3221 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3222 be caught using IGNORE-ERRORS.
3223 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3224 correctly when transforming recursive LABELS functions to LETs.
3225 (thanks to Alexey Dejneka)
3226 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3227 that are names of constants or global variables.
3228 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3229 alien routines with docstrings.
3230 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3231 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3233 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3234 (thanks to Eric Marsden)
3235 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3236 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3237 * Bug fix: The SPARC backend can now compile functions involving
3238 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3239 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3240 to a functions expecting a PATHNAME-DESIGNATOR.
3241 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3242 lambda lists. (thanks to Alexey Dejneka)
3243 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3244 functions so that EXPORTing the name of the class doesn't cause
3245 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3246 to Antonio Martinez for reporting this)
3247 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3248 constant-folded without failing an assertion. (thanks to Einar
3249 Floystad Dorum for reporting this)
3250 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3251 types are now dealt with more correctly by the compiler.
3252 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3253 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3254 behaviour analogous to RENAME-FILE. This puts its behaviour more
3255 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3257 * The fasl file version number has changed again. (because of the
3258 bug fix involving the names of PCL MAKE-INSTANCE functions)
3260 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3261 * bug fix: Floating point exceptions are treated much more
3262 consistently on the x86/Linux and PPC/Linux platforms.
3263 * Array initialization with :INITIAL-ELEMENT is now much faster for
3264 cases when the compiler cannot open code the array creation, but
3265 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3266 array accesses have also seen a speed increase.
3267 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3268 specified. (This was at the root of some bad interactions between
3269 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3271 * bug fix: Internal error arguments for undefined functions are now
3272 computed correctly on the PPC/Linux platform.
3273 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3274 Raymond Toy's patch for CMU CL)
3275 * Support for the Solaris 9 operating environment has been included
3276 (thanks to Daniel Merritt)
3277 * A very ugly but hopefully complete draft of the missing FFI chapter
3278 of the manual has been created by reformatting the corresponding
3279 CMU CL manual chapter into (currently very ugly and incoherent)
3280 DocBook and bringing it up to date for SBCL behavior. Thus, the
3281 manual is now essentially complete, at least by my extreme
3282 once-and-only-once standards, whereby it's acceptable to refer to
3283 the doc strings of SB-EXT functions as the primary documentation.
3284 * The fasl file version number has changed again, due to cleanup of
3285 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3286 is no longer a static symbol.)
3288 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3289 * SBCL now builds with OpenMCL (version 0.12) as the
3290 cross-compilation host; also, more progress has been made toward
3291 bootstrapping under CLISP.
3292 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3294 * bug 158 fixed: The compiler can now deal with integer loop
3295 increments different from 1; fixing this turned out also to fix
3297 * bug 169 fixed: no more bogus warnings about using lexical bindings
3298 despite the presence of perfectly good SPECIAL declarations (thanks
3299 to David Lichteblau)
3300 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3301 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3302 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3304 * bug 180 fixed: Method combination specifications no longer ignore
3305 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3306 * bug fix: Structure type predicate functions now check their argument
3307 count as they should.
3308 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3309 correctly. (thanks to Pierre Mai)
3310 * minor incompatible change: The --noprogrammer option is deprecated
3311 in favor of the new --disable-debugger option, which is very similar.
3312 (The major difference is that it takes effect at a slightly different
3313 time at startup, causing handling of errors in --sysinit and
3314 --userinit files will be affected differently.) The
3315 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3316 been added to allow this functionality to be controlled from ordinary
3317 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3318 maintainers, who might want to run non-interactive scripts to
3319 build SBCL cores which will later be used interactively.)
3320 * minor incompatible change: The LOAD function no longer, when given
3321 a wild pathname to load, loads all files matching that pathname.
3322 Instead, an error of type FILE-ERROR is signalled.
3324 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3325 * bug 147 fixed: The compiler preserves its block link/count
3326 invariants more correctly now so that it doesn't crash. (thanks
3328 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3330 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3331 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3332 no-value-for-this-character as NIL (as specified by ANSI).
3333 * HOST-NAMESTRING on physical pathnames now returns a string that is
3334 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3335 (thanks to Christophe Rhodes)
3336 * The Alpha port handles icache flushing more correctly. (thanks to
3338 * More progress has been made toward bootstrapping under CLISP. (thanks
3339 to Christophe Rhodes)
3340 * The fasl file format has changed again, because dynamic loading
3341 on OpenBSD (which has non-ELF object files) motivated some cleanups
3342 in the way that foreign symbols are transformed and passed around.
3343 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3344 is no longer treated as whitespace by the reader, but instead as
3345 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3346 |A
\7fB|, instead of A as it used to.
3348 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3349 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3350 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3351 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3352 this as of 0.7.1.45, but was left out of the previous news section
3353 (thanks to Dan Barlow)
3354 * SBCL now runs on the Solaris operating system on SPARC architectures
3355 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3356 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3357 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3358 * SPARC backend cleanups, allowing builds of cores optimized for V8
3359 and V9 SPARCS, and also emission of code targeted to a particular
3360 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3362 * SBCL is closer to bootstrapping under CLISP, thanks to various
3363 fixes by Christophe Rhodes.
3364 * The fasl file format has changed again, to allow the compiler's
3365 INFO database to support symbol macros.
3366 * The user manual (in doc/) is formatted into HTML more nicely.
3367 (thanks to coreythomas)
3368 * The system is smarter about SUBTYPEP relationships, especially
3369 those involving NOT types (including types such as ATOM which are
3370 represented internally using NOT types). Thus SUBTYPEP is less
3371 likely to return (VALUES NIL NIL) in general, and in particular
3372 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3374 * The fasl file format has changed again, because the internal
3375 representation of types now includes a new slot to support the new
3376 SUBTYPEP-of-NOT-types logic.
3377 * (not a change in the main branch of SBCL, but a related prototype
3378 which can hopefully be merged into the main branch of SBCL in the
3379 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3380 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3381 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3382 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3383 dispatch macro character combinations. (thanks to Alexey Dejneka)
3384 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3385 that superficially look like logical namestrings correctly.
3386 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3388 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3389 default supplied-p) for &optional and &key arguments. (thanks to
3392 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3393 * incompatible change: The compiler is now less aggressive about
3394 tail call optimization, doing it only when (> SPACE DEBUG) or
3395 (> SPEED DEBUG). (This is an incompatible change because there are
3396 programs which relied on the old CMU-CL-style behavior to optimize
3397 away their unbounded recursion which will now die of stack overflow.)
3398 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3399 for non-GENCGC systems has been increased to 20M (since that
3400 seems much closer to the likely performance optimum for modern
3401 systems than the old 4M value was)
3402 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3403 and *DEBUG-PRINT-LEVEL*
3404 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3405 of CMU CL's support for SPARC, and various endianness and other
3406 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3407 * new syntactic sugar for the Unix command line: --load foo.bar is now
3408 an alternate notation for --eval '(load "foo.bar")'.
3410 ** The system now detects stack overflow and handles it gracefully,
3411 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3412 optimization settings. (This is a good thing in general, and
3413 its introduction in this version should be particularly timely
3414 for anyone whose code fails because of suppression of tail
3416 ** The system now hunts for the C variable "environ" in a more
3417 devious way, to avoid segfaults when the C library version differs
3418 between compile time and run time. (thanks to Christophe Rhodes)
3419 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3420 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3421 ** The compiler no longer issues bogus style warnings for undefined
3422 classes in the same source file as the DEFCLASSes which defined
3423 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3425 ** fixes in CONDITION class precedence list for undefined function
3426 errors (thanks to Alexei Dejneka)
3427 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3428 correctly. (thanks to Dan Barlow)
3429 ** portability fixes aiming at bootstrapping under CLISP (thanks
3430 to Dave McDonald and Christophe Rhodes)
3431 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3432 ** fixes in type translation and and type inference (thanks to
3434 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3435 ** various fixes in the new ports (thanks to Dan Barlow)
3436 * several changes related to debugging:
3437 ** suppression of tail recursion, as noted above
3438 ** stack overflow detection, as noted above
3439 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3440 is now the default. (For some time encapsulation has been more
3441 reliable than the breakpoint-based :ENCAPSULATE NIL
3442 implementation, at least on X86 systems; and I just noticed that
3443 encapsulation also seems closer to the spirit of the ANSI
3446 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3448 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3449 up properly again. (There was a packaging bug in 0.7.0 which
3450 left their definitions in SB-SYS::LOAD-FOREIGN and
3451 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3452 vital for most things which interface to C-level interfaces,
3453 like extensions working with sockets or databases or
3454 Perl-compatible regexes or whatever, and the need to fix
3455 this bug is the main reason that 0.7.1 was released so
3457 ** DEFGENERIC is now choosier about the methods it redefines, so that
3458 reLOADing a previously-LOADed file containing DEFGENERICs does
3459 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3460 cycle now works as it should. (thanks to Alexey Dejneka)
3461 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3462 fixed by Christophe Rhodes. (He actually submitted this patch
3463 months ago, and I delayed until after 0.7.0.)
3464 ** Bug 111 (internal compiler confusion about runtime checks on
3465 FUNCTION types) was fixed by Alexey Dejneka.
3466 * Some internal cleanups (getting rid of variables which aren't
3467 needed now that the byte interpreter is gone) caused the fasl
3468 file format number to change again.
3470 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3471 * major incompatible change: The default fasl file extension, i.e. the
3472 default extension for files produced by COMPILE-FILE, has changed
3473 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3475 ** There are many changes in the implementation of the compiler.
3476 SBCL is now essentially a compiler-only implementation of ANSI
3477 Common Lisp. EVAL still "interprets" a few special cases, but
3478 almost all the interesting cases are handled by creating
3479 a LAMBDA expression, calling COMPILE on it, then calling
3480 FUNCALL on the result.
3481 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3482 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3483 Since the code is newer, there might still be some new bugs
3484 (though not as many as before Martin Atzmueller's fixes:-). But
3485 the new code is substantially simpler and clearer, and hopefully
3486 any remaining bugs will be simpler, less fundamental, and more
3487 fixable then the bugs in the old code.
3488 ** The revised compiler is still a little unsteady on its feet.
3490 *** The debugging information it produces (particularly the names
3491 of FUNCTION objects) is sometimes much less useful than what
3492 the old compiler produced.
3493 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3494 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3495 within a MACROLET) has been temporarily weakened.
3496 ** There are new compiler optimizations for various functions:
3497 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3498 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3499 *** the math functions TRUNCATE, FLOOR, and CEILING
3500 *** the function-of-all-trades COERCE
3501 Mostly these should be transparent, but there's one
3502 potentially-annoying problem (bug 117): when the compiler
3503 inline-expands a function and does type analysis on the result,
3504 it can create control paths which have type mismatches, and
3505 when it can't prove that those control paths aren't taken,
3506 it will issue WARNINGs about the type mismatches. This is
3507 a particular problem in practice for the new sequence functions.
3508 It's not clear how this should be fixed, and for now, a
3509 workaround is given in the entry for 117 in the BUGS file.
3510 ** (Because of the interaction between the two previous items --
3511 occasional inlining problems and new inline expansions -- some
3512 of the new sequence function optimizations won't really kick in
3513 completely until debugging information, and then inlining, are
3514 straightened out in some future version.)
3515 * minor incompatible changes:
3516 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3517 DIRECTORY no longer implicitly promotes NIL slots of its
3518 pathname argument to :WILD. In particular, when you ask for the
3519 contents of a directory (which you used to be able to do without
3520 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3521 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3522 ** changes in behavior that ANSI explicitly defines to be
3523 implementation dependent:
3524 *** The new compiler-only implementation still conforms with ANSI,
3525 but acts a little different than before. Besides the obvious
3526 changes in performance tradeoffs (that the cost per form passed
3527 to EVAL has gone up, and the cost per form executed by EVAL
3528 has gone down), the behavior of the system changes a little
3529 because there are no longer any interpreted function objects.
3530 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3531 e.g. doing COMPILE on the output of interactive DEFUN is
3533 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3535 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3536 has changed from (:CL) to NIL.
3537 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3538 e.g. #\Newline and #\Tab, don't have names specified in the
3539 ANSI Common Lisp standard, is now based on their ASCII symbolic
3540 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3541 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3542 are no longer used for output.
3543 ** changes in internal implementation constants:
3544 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3545 4 million. (If your application spends a lot of time GCing and
3546 you have a lot of RAM, you might want to experiment with
3547 increasing it even more.)
3548 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3549 However, almost all old code should still continue to work without
3550 immediate update, as SB-C-CALL is now a (deprecated) nickname
3552 ** Old operator names in the style DEF-FOO are now deprecated in
3553 favor of new corresponding names DEFINE-FOO, for consistency with
3554 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3555 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3556 internal symbols, but a few supported extensions like
3557 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3558 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3559 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3560 etc. as you get deeper into recursive calls to the debugger
3561 command loop, instead of the old "5]", "5]]", "5]]]"
3562 sequence. (I was motivated to do this when squabbles between
3563 ILISP and SBCL left me very deeply nested in the debugger. In the
3564 short term, this change will probably provoke more ILISP/SBCL
3565 squabbles, but hopefully it will be an improvement in the long run.)
3566 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3567 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3568 function, since declaiming return types involving aliens is
3569 (1) annoyingly messy to do by hand and (2) vital to efficient
3570 compilation of code which calls such functions.
3571 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3572 longer reexported by the SB-EXT package. They're solely useful
3573 for alien code, so it seems more logical that you should get
3574 them from the SB-ALIEN package, not in SB-EXT.
3575 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3576 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3577 features. Instead, the code that they used to control is always
3578 built into the system.
3579 * many other bug fixes
3580 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3581 advantage of the new EVAL-WHEN stuff and to clean them up in
3582 general, and they are now more ANSI-compliant in a number of
3583 ways. Martin Atzmueller is responsible for a lot of this.
3584 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3586 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3587 *** correct ERROR type for various file operations
3588 *** some fixes for Lisp streams
3589 *** DEFMETHOD syntax checking
3590 *** changing old weird representation of debug information as
3591 strings (which, among their other deficiencies, don't transform
3592 correctly when you rename packages, and don't change their
3593 print representation when you change things like *PACKAGE*
3594 and *PRINT-LENGTH*) to symbols and lists of symbols
3595 He also made several improvements and fixed several bugs in DESCRIBE.
3596 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3598 *** misbehavior of WRITE-STRING/WRITE-LINE
3599 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3600 and several other LOOP problems as well
3601 *** DIRECTORY when similar filenames are present
3602 *** DEFGENERIC with :METHOD options
3603 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3604 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3605 *** argument ordering in FIND with :TEST option
3606 *** mishandled package designator argument in APROPOS-LIST
3607 *** various problems in the backquote readmacro
3608 *** a bug in APROPOS
3609 *** probably some others that I'm not describing very well here,
3610 since the CVS log documents them by reference to sbcl-devel
3611 messages, and the SourceForge archives aren't working well.:-(
3612 ** Dan Barlow improved the Alpha port (and is making progress on the
3613 PPC port, for those of you who think different).
3614 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3615 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3616 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3617 continues to work on the SPARC port (for those of you in a position
3618 to look down upon our little PC-compatible boxes from a great height).
3619 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3620 and uses that copy, rather than the current dynamic value, when
3621 it's trying to decide whether to truncate output. Thus e.g.
3622 (let ((*print-lines* 50))
3623 (pprint-logical-block (stream nil)
3625 (let ((*print-lines* 8))
3626 (print (aref possiblybigthings i) stream)))))
3627 should now truncate the logical block only at 50 lines, instead of
3628 often truncating it at 8 lines, as it did before.
3629 * The doc/cmucl/ directory, containing old CMU CL documentation
3630 from the time of the fork, is no longer part of the base system.
3631 SourceForge has shut down its anonymous FTP service, and with it
3632 my original plan for distributing the old CMU CL documentation
3633 there. For now, if you need these files you can download an old
3634 SBCL source release and extract them from it.
3635 * The fasl file version number changed again, for dozens of reasons,
3636 some of which are apparent above.
3638 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3639 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3640 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3641 patch, so that DISASSEMBLE output is much nicer.
3642 * The code in the SB-PROFILE package now seems reasonably stable.
3643 I still haven't decided what the final interface should look like
3644 (I'd like PROFILE to interact cleanly with TRACE, since both
3645 facilities use function encapsulation) but if you have a need
3646 for profiling now, you can probably use it successfully with
3647 the current CMU-CL-style interface.
3648 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3649 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3650 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3651 link now returns the dangling link itself, and for similar
3652 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3653 link itself. (In these cases the old code signalled an error and
3654 looped endlessly, respectively.) Thus, DIRECTORY now works even
3655 in the presence of dangling and cyclic symbolic links.
3656 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3657 is now a supported extension again, since the consensus on
3658 sbcl-devel was that it can be useful for ordinary development
3659 work, not just for debugging SBCL itself.
3660 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3661 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3662 that currently-defined functions might be redefined later with
3663 different return types.
3664 * Hash tables can be printed readably, as inspired by CMU CL code
3665 of Eric Marsden and SBCL code of Martin Atzmueller.
3666 * better error handling in CLOS method combination, thanks to
3667 Martin Atzmueller porting Pierre Mai's CMU CL patches
3668 * more overflow fixes for >16Mbyte I/O buffers
3669 * A bug in READ has been fixed, so that now a single Ctrl-D
3670 character suffices to cause end-of-file on character streams.
3671 In particular, now you only need one Ctrl-D at the command
3672 line (not two) to exit SBCL.
3673 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3675 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3676 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3677 does the right thing.
3678 * The implementation of some type tests, especially for CONDITION
3679 types, is now tidier and maybe faster, due to CMU CL code
3680 originally by Douglas Crosher, ported by Martin Atzmueller.
3681 * Some math functions have been fixed, and there are new
3682 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3683 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3684 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3685 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3686 likely to be built into sbcl-0.7.0.)
3687 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3688 CL, ported to SBCL by Martin Atzmueller)
3689 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3690 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3691 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3692 * There's a new slam.sh hack to shorten the edit/compile/debug
3693 cycle for low-level changes to SBCL itself, and a new
3694 :SB-AFTER-XC-CORE target feature to control the generation of
3695 the after-xc.core file needed by slam.sh.
3696 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3697 COMPILE-FILE is no longer supported, so that now every function
3698 gets an entry point, so that block compilation looks a little
3699 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3700 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3701 returns the number of bytes consed since the system started,
3702 rather than the number consed since the first time the function
3703 was called. (The new definition parallels ANSI functions like
3704 CL:GET-INTERNAL-RUN-TIME.)
3705 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3706 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3707 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3708 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3709 * Fasl file version numbers are now independent of the target CPU,
3710 since historically most system changes which required version
3711 number changes have affected all CPUs equally. Similarly,
3712 the byte fasl file version is now equal to the ordinary
3715 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3716 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3717 is no longer recognized. I apologize for this, because it was
3718 listed in SB-EXT as a supported extension, but I found that
3719 its existing behavior was poorly specified, as well as incorrectly
3720 specified, and it looked like too much of a mess to straighten it
3721 out. I have enough on my hands trying to get ANSI stuff to work..
3722 * many patches ported from CMU CL by Martin Atzmueller, with
3723 half a dozen bug fixes in pretty-printing and the debugger, and
3724 half a dozen others elsewhere
3725 * fixed bug 13: Floating point infinities are now supported again.
3726 They might still be a little bit flaky, but thanks to bug reports
3727 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3728 as flaky as they were.
3729 * The --noprogrammer command line option is now supported. (Its
3730 behavior is slightly different in detail from what the old man
3731 page claimed it would do, but it's still appropriate under the
3732 same circumstances that the man page talks about.)
3733 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3734 are now supported, and enabled by default. Thus, the compiler can
3735 handle many floating point and complex operations much less
3736 inefficiently. (Thus e.g. you can implement a complex FFT
3738 * The compiler now detects type mismatches between DECLAIM FTYPE
3739 and DEFUN better, and implements CHECK-TYPE more correctly, and
3740 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3741 thanks to patches from Martin Atzmueller.
3742 * various fixes to make the cross-compiler more portable to
3743 ANSI-conforming-but-different cross-compilation hosts (notably
3744 Lispworks for Windows, following bug reports from Arthur Lemmens)
3745 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3746 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3747 fixed thanks to Pierre Mai's CMU CL patches.
3748 * fixes to keep the system from overflowing internal counters when
3749 it tries to use i/o buffers larger than 16M bytes
3750 * fixed bug 45a: Various internal functions required to support
3751 complex special functions have been merged from CMU CL sources.
3752 (When I was first setting up SBCL, I misunderstood a compile-time
3753 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3754 * improved support for type intersection and union, fixing bug 12
3755 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3756 more obscure bugs as well
3757 * some steps toward byte-compiling non-performance-critical
3758 parts of the system, courtesy of patches from Martin Atzmueller
3759 * Christophe Rhodes has made some debian packages of sbcl at
3760 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3761 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3762 stable, but are nonetheless usable. When he's ready, I'd be happy
3763 to add them to the SourceForge "File Releases" section. (And if
3764 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3765 * new fasl file format version number (because of changes in
3766 internal representation of (OR ..) types to accommodate the new
3767 support for (AND ..) types, among other things)
3769 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3770 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3772 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3774 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3775 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3776 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3777 * DESCRIBE now works on structure objects again.
3778 * Most function call argument type mismatches are now handled as
3779 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3780 know whether the function will be redefined before the call is
3781 executed. (The compiler could flag local calls with full WARNINGs,
3782 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3783 it doesn't keep track of enough information to know whether calls
3784 are local in this sense.)
3785 * Compiler output is now more verbose, with messages truncated
3786 later than before. (There should be some supported way for users
3787 to override the default verbosity, but I haven't decided how to
3788 provide it yet, so this behavior is still controlled by the internal
3789 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3790 src/compiler/ir1util.lisp.)
3791 * Fasl file format version numbers have increased again, because
3792 support for the Gray streams extension changes the layout of the
3793 system's STREAM objects.
3794 * The Gray subclassable streams extension now works, thanks to a
3795 patch from Martin Atzmueller.
3796 * The full LOAD-FOREIGN extension (not just the primitive
3797 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3798 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3799 but like most other programs, it defaults to copying the Unix
3800 environment from the original process instead of starting the
3801 new process in an empty environment.
3802 * Extensions which manipulate the Unix environment now support
3803 an :ENVIRONMENT keyword option which doesn't smash case or
3804 do other bad things. The CMU-CL-style :ENV option is retained
3805 for porting convenience.
3806 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3807 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3809 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3811 * A patch from Martin Atzmueller seems to have solved the SIGINT
3812 problem, and as far as we know, signal-handling now works cleanly.
3813 (If you find any new bugs, please report them!)
3814 * The system no longer defaults Lisp source file names to types
3815 ".l", ".cl", or ".lsp", but only to ".lisp".
3816 * The compiler no longer uses special default file extensions for
3817 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3818 seems to expect a single default extension for all compiled code,
3819 and there's no compelling reason to try to stretch the standard
3820 to allow two different extensions.) Instead, byte-compiled files
3821 default to the same extension as native-compiled files.
3822 * Fasl file format version numbers have increased again, because
3823 a rearrangement of internal implementation packages made some
3824 dumped symbols in old fasl files unreadable in new cores.
3825 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3826 many fewer weird special cases.
3827 * Bug #17 (differing COMPILE-FILE behavior between logical and
3828 physical pathnames) has been fixed, and some related misbehavior too,
3829 thanks to a patch from Martin Atzmueller.
3830 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3831 by Tim Moore, ported to SBCL by Martin Atzmueller.
3832 * Martin Atzmueller fixed several filesystem-related problems,
3833 including bug #36, in part by porting CMU CL patches, which were
3834 written in part by Paul Werkowski.
3835 * More compiler warnings in src/runtime/ are gone, thanks to
3836 more patches from Martin Atzmueller.
3837 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3840 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3842 * DESCRIBE now works on CONDITION objects.
3843 * The debugger now handles errors which arise when trying to print
3844 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3846 * The build system now uses an additional file, customize-target-features.lisp,
3847 to allow local modifications to the target *FEATURES* list. (The point of
3848 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3849 debugging features enabled, without having to worry about propagating it
3850 into everyone's system when I do a "cvs update".) When no
3851 customize-target-features.lisp file exists, the target *FEATURES* list
3852 should be constructed the same way as before.
3853 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3854 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3855 warn about behavior which is undefined under the ANSI spec. Note: This
3856 is specified by ANSI, but it's not very popular with programmers.
3857 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3858 macro in the SBCL sources for an example of a workaround which you
3859 might use to make portable ANSI-standard code which does what you want.
3860 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3861 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3862 when it's used as a non-toplevel form. (This is required in order
3863 to implement the DEFCONSTANT-EQX macro.)
3864 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3865 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3866 funny things, due to bugs in the implementation of EVAL-WHEN
3867 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3868 non-toplevel DEFCONSTANTs are uncommon.)
3869 * The core file version number and fasl file version number have been
3870 incremented, because the old noncompliant DEFCONSTANT behavior involved
3871 calling functions which no longer exist, and because I also took the
3872 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3873 * fixed bug 1 (error handling before read-eval-print loop starts), and
3874 redid debugger restarts and related debugger commands somewhat while
3876 ** The QUIT debugger command is gone, since it did something
3877 rather different than the SB-EXT:QUIT command, and since it never
3878 worked properly outside the main toplevel read/eval/print loop.
3879 Invoking the new TOPLEVEL restart provides the same functionality.
3880 ** The GO debugger command is also gone, since you can just invoke
3881 the CONTINUE restart directly instead.
3882 ** The TOP debugger command is also gone, since it's redundant with the
3883 FRAME 0 command, and since it interfered with abbreviations for the
3885 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3887 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3888 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3889 some porting work by Martin Atzmueller)
3890 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3891 with a boatload of patches which he ported to SBCL. Now that those
3893 ** The system tries to make sure that its low-priority messages
3894 are prefixed by semicolons, to help people who like to use
3895 syntax highlighting in their ILISP buffer. (This patch
3896 was originally due to Raymond Toy.)
3897 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3898 patches originally written by Raymond Toy.
3899 ** The compiler understands coercion between single-value and
3900 multiple-VALUES type expressions better, getting rid of some very
3901 weird behavior, thanks to patches originally by Robert MacLachlan
3902 and Douglas Crosher.
3903 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3904 lambda lists, thanks to a patch originally by Pierre Mai.
3905 ** The system no longer bogusly warns about "abbreviated type
3907 ** The compiler gets less confused by inlining and RETURN-FROM,
3908 thanks to some patches originally by Tim Moore.
3909 ** The system no longer hangs when dumping circular lists to fasl
3910 files, thanks to a patch originally from Douglas Crosher.
3911 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3912 undefined function error.
3913 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3914 in structure slot initforms) for now, documented workaround instead:-|
3915 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3916 * fixed bug 5: added stubs for various Gray stream functions called
3917 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3918 installed, at least appropriate type errors are generated
3919 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3920 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3921 consistently in DEFMETHOD forms.
3922 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3923 it doesn't seem to affect SBCL after all
3924 * The C runtime system now builds with better optimization and many
3925 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3927 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3929 * The system is now under CVS at SourceForge (instead of the
3930 CVS repository on my home machine).
3931 * The new signal handling code has been tweaked to treat register
3932 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3933 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3934 downstream, e.g. in debug-int.lisp, has implicit dependencies
3935 on the unsignedness of integer representation of machine words,
3936 and that caused the system to bomb out with infinite regress
3937 when trying to recover from type errors involving signed values,
3938 e.g. (BUTLAST '(1 2 3) -1).)
3939 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3940 (This was one of the bugs Peter Van Eynde reported back in July.)
3941 * The system now uses code inspired by Colin Walters' O(N)
3942 implementation of MAP (from the cmucl-imp@cons.org mailing
3943 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3944 inline the MAP operation, and there is more than one
3945 sequence argument to the MAP call (so that it can't just
3946 do ETYPECASE once and for all based on the type of the
3947 single sequence argument). (The old non-inline implementation
3948 of the general M-argument sequence-of-length-N case required
3949 O(M*N*N) time when any of the sequence arguments were LISTs.)
3950 * The QUIT :UNIX-CODE keyword argument has been renamed to
3951 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3952 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3954 * Raymond Wiker's patches to port dynamic loading from Linux to
3955 FreeBSD have been added.
3956 * The BUGS file is now more nearly up to date, thanks in large part
3957 to Martin Atzmueller's review of it.
3958 * The debugger now flushes standard output streams before it begins
3959 its output ("debugger invoked" and so forth).
3960 * The core version number and fasl file version number have both
3961 been incremented, because of incompatible changes in the layout
3963 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3964 used to not be called for a saved Lisp image.)
3965 * Martin Atzmueller's version of a patch to fix a compiler crash,
3966 as posted on sbcl-devel 13 September 2000, has been installed.
3967 * Instead of installing Martin Atzmueller's patch for the
3968 compiler transform for SUBSEQ, I deleted the compiler transform,
3969 and transforms for some similar consing operations.
3970 * A bug in signal handling which kept TRACE from working on OpenBSD
3972 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3973 away by constant folding
3974 * The system now defines its address space constants in one place
3975 (in the Lisp sources), and propagates them automatically elsewhere
3976 (through GENESIS and the sbcl.h file). Therefore, patching the
3977 address map is less unnecessarily tedious and error-prone. The
3978 Lisp names of address space constants have also been systematized.
3979 * CVS tags like dollar-Header-dollar have been removed from
3980 the sources, because they have never saved me trouble and
3981 they've been source of trouble working with patches and other
3982 diff-related operations.
3983 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3984 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3986 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3988 * The system has been ported to OpenBSD.
3989 * The system now compiles with a simple "sh make.sh" on the systems
3990 that it's supported on. I.e., now you no longer need to tweak
3991 text in the target-features.lisp-expr and symlinks in src/runtime/
3992 by hand, the make.sh takes care of it for you.
3993 * The system is no longer so grossly inefficient when compiling code
3994 involving vectors implemented as general (not simple) vectors (VECTOR T),
3995 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3996 which dares to use the various sequence functions on non-simple
3997 vectors, takes less of a performance hit.
3998 * There is now a primitive type predicate VECTOR-T-P
3999 to test for the (VECTOR T) type, so that e.g.
4000 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4001 can now be compiled with some semblance of efficiency. (The old code
4002 turned the type declaration into a full call to %TYPEP at runtime!)
4003 * AREF on (VECTOR T) is still not fast, since it's still compiled
4004 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4005 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4006 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4007 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4008 without making full calls to SUBTYPEP at runtime.
4009 (Some analogous efficiency issues for non-simple vectors specialized to
4010 element types other than T, or for non-simple multidimensional arrays,
4011 have not been addressed. They could almost certainly be handled the
4012 same way if anyone is motivated to do so.)
4013 * The changes in array handling break binary compatibility, so
4014 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4015 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4016 instead of (VALUES T T).
4017 * By following the instructions that Dan Barlow posted to sbcl-devel
4018 on 2 July 2000, I was able to enable primitive dynamic object
4019 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4020 functionality is not implemented (since it calls ld to resolve
4021 library references automatically, requiring RUN-PROGRAM for its
4022 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4023 try to resolve library references) is now supported.
4024 * The system now flushes the standard output streams when it terminates,
4025 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4026 them at several other probably-convenient times, e.g. in each pass of
4027 the toplevel read-eval-print loop, and after evaluating a form given
4028 as an "--eval" command-line option. (These changes were motivated by a
4029 discussion of stream flushing issues on cmucl-imp in August 2000.)
4030 * The source transform for TYPEP of array types no longer assumes
4031 that an array whose element type is a not-yet-defined type
4032 is implemented as an array of T, but instead punts, so that the
4033 type will be interpreted at runtime.
4034 * There is now some support for cross-compiling in make.sh: each of
4035 the phases of make.sh has its own script. (This should be transparent
4036 to people doing ordinary, non-cross-compile builds.)
4037 * Since my laptop doesn't have hundreds of megabytes of memory like
4038 my desktop machine, I became more motivated to do some items on
4039 my to-do list in order to reduce the size of the system a little:
4040 ** Arrange for various needed-only-at-cold-init things to be
4041 uninterned after cold init. To support this, those things have
4042 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4043 symbols with such names are now uninterned after cold init).
4044 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4045 fasl files for cold load.
4046 ** Remove the old compiler structure pooling code (which used to
4047 be conditional on the target feature :SB-ALLOC) completely.
4048 ** Redo the representation of some data in cold init to be more compact.
4049 (I also looked into supporting byte compiled code at bootstrap time,
4050 which would probably reduce the size of the system a lot, but that
4051 looked too complicated, so I punted for now.)
4052 * The maximum signal nesting depth in the src/runtime/ support code has
4053 been reduced from 4096 to 256. (I don't know any reason for the very
4054 large old value. If the new smaller value turns out to break something,
4055 I'll probably just bump it back up.)
4056 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4058 * Many, many bugs reported by Peter Van Eynde have been added to
4059 the BUGS list; some have even been fixed.
4060 * While enabling dynamic object file loading, I tried to make the
4061 code easier to understand, renaming various functions and variables
4062 with less ambiguous names, and changing some function calling
4063 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4064 * While trying to figure out how to do the OpenBSD port, I tried to
4065 clean up some of the code in src/runtime/. In particular, I dropped
4066 support for non-POSIX signal handling, added various comments,
4067 tweaked the code to reduce the number of compilation warnings, and
4068 renamed some files to increase consistency.
4069 * To support the new automatic configuration functionality in make.sh,
4070 the source file target-features.lisp-expr has been replaced with the
4071 source file base-target-features.lisp-expr and the machine-generated
4072 file local-target-features.lisp-expr.
4073 * fixed a stupid quoting error in make.sh so that using CMU CL
4074 "lisp -batch" as cross-compilation host works again
4076 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4078 * DESCRIBE no longer tries to call itself recursively to describe
4079 bound/fbound values, so that it no longer fails on symbols which are
4080 bound to themselves (like keywords, T, and NIL).
4081 * DESCRIBE now works on generic functions.
4082 * The printer now prints less-screwed-up representations of closures
4083 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4084 * A private symbol is used instead of the :EMPTY keyword previously
4085 used to mark empty slots in hash tables. Thus
4086 (DEFVAR *HT* (MAKE-HASH-TABLE))
4087 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4088 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4089 now does what ANSI says that it should. (You can still get
4090 similar noncompliant behavior if bang on the hash table
4091 implementation with all the symbols you get back from
4092 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4093 This breaks binary compatibility, since tests for equality to
4094 :EMPTY are wired into things like the macroexpansion of
4095 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4097 * There's now a minimal placeholder implementation for CL:STEP,
4098 as required by ANSI.
4099 * An obscure bug in the interaction of the normal compiler, the byte
4100 compiler, inlining, and structure predicates has been patched
4101 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4102 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4104 * Missing ordinary arguments in a macro call are now detected even
4105 when the macro lambda list contains &KEY or &REST.
4106 * The debugger no longer complains about encountering the top of the
4107 stack when you type "FRAME 0" to explicitly instruct it to go to
4108 the top of the stack. And it now prints the frame you request even
4109 if it's the current frame (instead of saying "You are here.").
4110 * As specified by ANSI, the system now always prints keywords
4111 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4113 * The default initial SIZE of HASH-TABLEs is now smaller.
4114 * Type information from CLOS class dispatch is now propagated
4115 into DEFMETHOD bodies, so that e.g.
4116 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4118 is now basically equivalent to
4119 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4120 (DECLARE (TYPE SINGLE-FLOAT X))
4122 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4123 operation, without having to do run-time type dispatch.
4124 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4125 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4126 * Since I don't seem to be making big file reorganizations very often
4127 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4128 is overflowing my ability to conveniently back them up), I've finally
4129 checked the system into CVS. (The CVS repository is on my home system,
4130 not at SourceForge -- putting it on SourceForge might come later.)
4131 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4132 high-level GC-NOTIFY-FOO functions send their output. (There's
4133 still very little control of where low-level verbose GC functions
4134 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4135 less than it used to -- the GC-NOTIFY-FOO functions are now under
4136 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4137 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4138 in only one place in the source code, and propagates it automatically
4139 everywhere that it's needed. Thus e.g. when I bump the version from
4140 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4141 * The C source files now include boilerplate legalese and documentation
4142 at the head of each file (just as the Lisp source files already did).
4143 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4144 to his page will be replaced with a link to his new CLiki service.
4146 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4148 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4149 * The build process now looks for GNU make under the default name "gmake",
4150 instead of "make" as it used to. If GNU make is not available as "gmake"
4151 on your system, you can change this default behavior by setting the
4152 GNUMAKE environment variable.
4153 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4154 can build without error under CMU CL.
4156 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4158 * There is now a partial SBCL user manual (with some new text and some
4159 text cribbed from the CMU CL manual).
4160 * The beginnings of a profiler have been added (starting with the
4161 CMU CL profiler and simplifying and cleaning up). Eventually the
4162 main interface should be through the TRACE macro, but for now,
4163 it's still accessed through vaguely CMU-CL-style functions and macros
4164 exported from the package SB-PROFILE.
4165 * Some problems left over from porting CMU CL to the new
4166 cross-compilation bootstrap process have been cleaned up:
4167 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4168 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4169 part when getting rid of the compiler *BACKEND* stuff.)
4170 ** The value of *NULL-TYPE* was screwed up, because it was
4171 being initialized before the type system knew the final
4172 definition of the 'NULL type. This screwed up several key
4173 optimizations in the compiler, causing inefficiency in all sorts
4174 of places. (I found it because I wanted to understand why
4175 GET-INTERNAL-RUN-TIME was consing.)
4176 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4177 PROCLAIM FTYPE information. Unfortunately this broke binary
4178 compatibility again, since now the forms output by DEFGENERIC
4179 to refer to functions which didn't exist in 0.6.3.
4180 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4181 can use the new (as of 0.6.3) transform for SOME into MAP into
4183 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4184 values to (INTEGER 0 1000000), so that the time code will no longer
4185 occasionally get blown up by Linux returning 1000000 microseconds
4186 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4187 its output conform to the ANSI spec. (Alas, this makes its output
4188 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4189 * A full call to MAP NIL with a single sequence argument no longer conses.
4190 * fixes to problems pointed out by Martin Atzmueller:
4191 * The manual page no longer talks about multiprocessing as though
4192 it were currently supported.
4193 * The ILISP support patches have been removed from the distribution,
4194 because as of version 5.10.1, ILISP now supports SBCL without us
4195 having to maintain patches.
4196 * added a modified version of Raymond Toy's recent CMU CL patch for
4197 EQUALP comparison of HASH-TABLE
4199 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4201 * The system still can't cross-compile itself with
4202 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4203 entails), but at least it can compile more of itself that way
4204 than it used to be able to, and various buglets which were uncovered
4205 by trying to cross-compile itself that way have now been fixed.
4206 * This release breaks binary compatibility again. This time
4207 at least I've incremented the FASL file format version to 2, so that the
4208 problem can be detected reliably instead of just causing weird errors.
4209 * various new style warnings:
4210 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4211 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4212 ** using the lexical binding of a variable named in the *FOO* style
4213 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4214 as specified by ANSI.
4215 * *RANDOM-STATE* is no longer automatically initialized from
4216 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4217 default behavior of the system is to repeat its behavior every time
4218 it's run. If you'd like to change this behavior, you can always
4219 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4220 old convention there was no comparably easy way to get the system to
4221 repeat its behavior every time it was run.
4222 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4223 FTYPE declarations has been removed, in favor of their ANSI
4224 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4225 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4226 the types of their sequence arguments can be determined at compile time.
4227 This is done through a new open code expansion for MAP which eliminates
4228 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4229 argument types can be determined at compile time.
4230 * The optimizer now transforms COERCE into an identity operation when it
4231 can prove that the coerced object is already of the correct type. (This
4232 can be a win for machine generated code, including the output of other
4233 optimization transforms, such as the MAP transform above.)
4234 * Credit information has been moved from source file headers into CREDITS.
4235 * Source file headers have been made more standard.
4236 * The CASE macro now compiles without complaining even when it has
4239 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4241 * (Note that the way that the PCL macroexpansions were rewritten
4242 to accommodate the change in DEFGENERIC below breaks binary
4243 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4244 not run under sbcl-0.6.2. Once we get out of alpha releases,
4245 i.e. hit release 1.0.0, we'll probably try to maintain binary
4246 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4247 and sbcl-1.4.4. Until then, however, it might be fairly common
4248 for maintenance releases to break binary compatibility.)
4249 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4251 * The reporting of unhandled signals has been changed to print some
4252 explanatory text as well as the report form. (Previously only
4253 the report form was printed.)
4254 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4255 it defines, so that the compiler no longer issues undefined function
4256 warnings for compiled-but-not-yet-loaded generic functions.
4257 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4258 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4259 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4260 section of the ANSI spec are supported.
4261 * The "" nickname for the "KEYWORD" package has been removed.
4262 The reader still handles symbol tokens which begin with a package marker
4263 as keywords, but it doesn't expose its mechanism for doing so in the
4264 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4265 * The system now issues STYLE-WARNINGs for contradictory TYPE
4266 proclamations. (Warnings for contradictory FTYPE proclamations would
4267 be nice too, but those can't be done usefully unless the type system
4268 is made smarter about FUNCTION types.)
4269 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4270 systematized, so that "*target-*.lisp is supposed to exist only on the
4271 target and imply that there's a related file which exists on the
4272 host, and *host-*.lisp is supposed to exist only on the host and imply
4273 that there's a related file which exists on the target. This involves a
4274 lot of renaming. Hopefully the acute confusion caused by the renaming
4275 will be justified by the reduction in chronic confusion..
4276 ** runtime-type.lisp -> early-target-type.lisp
4277 ** target-type.lisp -> late-target-type.lisp
4278 ** early-host-format.lisp -> early-format.lisp
4279 ** late-host-format.lisp -> late-format.lisp
4280 ** host-error.lisp -> misc-error.lisp
4281 ** early-error.lisp -> early-target-error.lisp
4282 ** late-error.lisp -> late-target-error.lisp
4283 ** host-defboot.lisp -> early-defboot.lisp
4284 ** code/misc.lisp -> code/target-misc.lisp
4285 ** code/host-misc.lisp -> code/misc.lisp
4286 ** code/numbers.lisp -> code/target-numbers.lisp
4287 ** code/early-numbers.lisp -> numbers.lisp
4288 ** early-host-type.lisp -> early-type.lisp
4289 ** late-host-type.lisp -> late-type.lisp
4290 ** host-typep.lisp -> typep.lisp
4291 ** load.lisp -> target-load.lisp
4292 ** host-load.lisp -> load.lisp
4293 ** host-disassem.lisp -> disassem.lisp
4294 ** host-insts.lisp -> insts.lisp
4295 ** byte-comp.lisp -> target-byte-comp.lisp
4296 ** host-byte-comp.lisp -> byte-comp.lisp
4297 ** host-signal.lisp -> signal.lisp
4298 ** host-defstruct.lisp -> defstruct.lisp
4299 ** late-target-type.lisp -> deftypes-for-target.lisp
4300 Furthermore, several other previously target-only files foo.lisp (e.g.
4301 hash-table.lisp and random.lisp) have been split into a target-and-host
4302 foo.lisp file and a target-only target-foo.lisp file, with their key type
4303 definitions in the target-and-host part, so that the cross-compiler will
4304 know more about target types.
4305 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4306 gone away. In their place are various *BACKEND-FOO* variables
4307 corresponding to the slots of the old structure.
4308 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4309 into the target SBCL has been fixed.
4310 * The chill.lisp system for loading cold code into a running SBCL
4312 * Support for the CMU CL "scavenger hook" extension has been removed.
4313 (It was undocumented and unused in the CMU CL sources that SBCL was
4314 derived from, and stale in sbcl-0.6.1.)
4315 * Various errors in the cross-compiler type system were detected
4316 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4317 (enabling various consistency checks). Many of them were fixed,
4318 but some hard problems remain, so the compiler is back to
4319 running without *TYPE-SYSTEM-INITIALIZED* for now.
4320 * As part of the cross-compiler type system cleanup, I implemented
4321 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4322 * I have started adding UNCROSS calls throughout the type system
4323 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4324 on cross-compiler input files will be able to go away, and various
4326 * CONSTANTP now returns true for quoted forms (as explicitly required
4329 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4331 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4332 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4333 (MAKE-LIST -1). In the longer term, it ought to become true
4334 that declarations are assertions even at SAFETY 1. For now, it's not
4335 quite true even at SAFETY 3, but it's at least more nearly true..
4336 (Note that this change seems to increases the size of the system by
4337 O(5%) and to decrease the speed of the compiler by 20% or more.)
4338 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4339 for the problem of printing dozens of lines of distracting information
4340 about low-level system machinery as part of the top stack frame
4341 on entry to the debugger when an undefined function was called.
4342 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4343 is not reset to COMMON-LISP-USER.
4344 * Compilation of stuff related to dyncount.lisp has been made conditional
4345 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4346 smaller. The various dyncount-related symbols have been moved into
4347 a new "SB-DYNCOUNT" package.
4348 * tty-inspect.lisp has been renamed to inspect.lisp.
4349 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4350 feature has gone away. (When we eventually port to other flavors of
4351 libc and/or Unix, we'll try to make the differences between flavors
4352 invisible at the user level.)
4353 * Various other *FEATURES* tags, and/or their associated conditionals,
4354 have been removed if obsolescent, or given better documentation, or
4355 sometimes given more-mnemonic names.
4357 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4359 * tidied up "make.sh" script
4360 * tidied up system directory structure
4361 * better "clean.sh" behavior
4362 * added doc/FOR-CMUCL-DEVELOPERS
4363 * many many small tweaks to output format, e.g. removing possibly-confusing
4364 trailing #\. character in DESCRIBE-INSTANCE
4365 * (EQUALP #\A 'A) no longer signals an error.
4366 * new hashing code, including EQUALP hashing
4367 * tidied up Lisp initialization and toplevel
4368 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4369 * command line argument processing
4370 * added POSIX-GETENV function to deal with Unix-ish environment variables
4371 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4372 terminating SBCL on EOF
4373 * non-verbose GC by default
4374 * There is no more "sbcl" shell script; the sbcl file is now the C
4375 runtime executable (just like CMU CL).
4376 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4378 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4379 symbol and package information as Lisp native symbol and package objects
4380 instead of strings naming symbols and strings naming packages. This way,
4381 whenever packages are renamed (as in warm init), debug information is
4382 transformed along with everything else.
4383 * tweaked the optimization policy declarations which control the building
4384 of SBCL itself. Now, among other things, the system no longer saves
4385 source location debugging information. (This helps two problems at once
4386 by reducing SBCL size and by keeping SBCL from trying to look for its
4387 sources -- which may not exist -- when reporting errors.)
4388 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4389 debugging and testing purposes
4390 * cleaned up printing, making the printer call PRINT-OBJECT for
4391 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4392 methods, giving nearly-ANSI behavior
4393 * converted almost all special variables to use *FOO* naming convention
4394 * deleted PARSE-TIME functionality, since it can be done portably
4395 * moved some files out of cold init into warm init
4396 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4398 * regularized formatting of source files
4399 * added an install.sh script
4400 * fixed ridiculous memory usage of cross-compiler by making
4401 compiler/alloc.lisp not try to do pooling unless it can hook
4402 itself into the GC of the cross-compilation host. Now the system
4403 builds nicely on my old laptop.
4404 * added :SB-ALLOC in target-features.lisp-expr
4405 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4406 was not implemented)
4407 * re-did condition handling and note reporting in the compiler. Notes
4408 are no longer handled by signalling conditions. Style warnings
4409 and warnings are handled more correctly and reported in such a way
4410 that it's easy to find one or the other in your output (so that you
4411 can e.g. figure out which of many problems caused COMPILE-FILE to
4413 * changed the severity of several compiler warnings from full WARNING
4414 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4415 compiler note reporting so that it doesn't use the condition system
4416 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4417 COMPILE-FILE command)
4418 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4419 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4420 no longer supported, since their implementation was deeply intertwingled
4421 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4422 * removed bogus "support" for compiler macros named (SETF FOO), and
4423 removed the compiler macro for SETF INFO (but only after making a fool
4424 of myself on the cmucl-imp mailing list by posting a bogus patch for
4425 DEFINE-COMPILER-MACRO..)
4426 * Compiled files containing forms which have side effects on the Lisp
4427 reader (such as DEFPACKAGE forms) are now handled more correctly.
4428 (Compiler queuing of top level lambdas has been suppressed by setting
4429 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4430 * deleted various currently-unused source files, e.g. gengc.lisp. They
4431 may be added back at some point e.g. when porting to other architectures,
4432 but until they are it's distracting to distribute them and to try to
4434 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4435 there were so many of them they're just distractions, and UNCROSS is
4436 known to be able to handle the current sources
4437 * moved PROFILE functionality into TRACE, so that it will be clear
4438 how the wrapping and unwrapping of functions when you profile them
4439 interacts with the wrapping and unwrapping of functions when you
4440 trace them. (Actually, the functionality isn't there yet, but at least
4441 the interface specification is there. Hopefully, the functionality will
4442 arrive with some future maintenance release.)
4443 * removed host-oops.lisp
4444 * changed signature of QUIT function to allow UNIX-CODE argument
4445 * fixed READ-SEQUENCE bug
4446 * tweaked verbose GC output so that it looks more like the progress
4447 output that ANSI specifies for functions like LOAD
4448 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4449 * added <http://sbcl.sourceforge.com> to the banner information printed by