1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
3 * enhancement: CONS can now stack-allocate on x86 and
4 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
6 * enhancement: nested lists can now be stack allocated on
7 platforms providing stack allocation support.
8 * enhancement: dynamic-extent support has been extended to support
9 cases where there are multiple possible sources for the stack
11 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
12 as specified by the CLHS. (thanks to Tobias Rittweiler)
14 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
15 * minor incompatible change: the MSI installer on Windows no longer
16 associates .lisp and .fasl files with the installed SBCL.
17 * minor incompatible change: :UNIX is no longer present in *FEATURES*
18 on Windows. (thanks to Luis Oliviera)
19 * new feature: SB-CLTL2 contrib module now implements
20 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
21 * optimization: scavenging weak pointers is now more efficient,
22 requiring O(1) instead of O(N) per weak pointer to identify
23 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
24 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
25 method bodies using constant slot-names when the first argument is
26 a specializer parameter for the method.
27 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
28 names, when the class of the instance is a direct instance
29 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
30 as slow as the constant slot-name case.)
31 * optimization: member type construction is now O(N) instead
33 * optimization: UNION and NUNION are now O(N+M) for large
34 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
36 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
37 now more readable in environments like Slime which display it.
38 (thanks to Tobias C. Rittweiler)
39 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
40 was able to infer the correct argument type for the object on which
41 the CAS operation was being performed.
42 * bug fix: copy propagation interfered with parallel assignment
43 semantics in local calls. (reported by Paul Khuong)
44 * bug fix: the signed modular fixnum shift compiled to wrong code on
45 x86 and x86-64. (spotted by a slight modification to some of
48 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
49 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
50 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
51 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
53 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
54 and generic function now print the object identity for anonymous
55 instances. (thanks to Attila Lendvai)
56 * enhancement: as an extension to MOP, SBCL now supports SETF of
57 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
58 (thanks to Attila Lendvai)
59 * enhancement: sb-cover has an alternate annotation mode which puts
60 the form annotations at the car of the form, instead of annotating
61 it completely. The benefit of this mode is that it shows better
62 which forms were instrumented by the compiler.
63 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
64 by printer control variables. (reported by Dan Corkill)
65 * bug fix: system leaked memory when delayed output was performed by
66 the OS in smaller chunks then expected. (thanks to David Smith)
67 * bug fix: system leaked memory when file streams were not closed
69 * bug fix: large objects written to slow streams that were modified
70 after the write could end up with the modified state written to
71 the underlying file descriptor.
72 * bug fix: multiple threads operating in parallel on the same stream
73 could cause buffer-overflows.
74 * bug fix: source location information is stored correctly
75 (broken since 1.0.6). This bug would generally show up as the
76 Slime debugger higlighting the wrong form when the sldb-show-source
78 * bug fix: Branch forms are again annotated as branches in the sb-cover
80 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
81 on x86-64. (reported by Gregory Vanuxem)
82 * bug fix: EQUALP could return wrong results for structures with raw
83 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
84 word). (reported by Vjacheslav Fyodorov)
85 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
88 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
89 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
90 atomic compare-and-swap operations on threaded platforms.
91 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
92 allows assining a global minimum value to optimization qualities
93 (overriding proclamations and declarations).
94 * enhancement: closed over variables can be stack-allocated on x86
96 * performance bug fix: GETHASH and (SETF GETHASH) are once again
98 * optimization: slot definition lookup is now O(1). This speeds up
99 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
100 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
101 * optimization: ASSOC and MEMBER can now be open-coded for all
102 combinations of keyword arguments when second argument is constant
103 and SPEED >= SPACE. In other cases a specialized version is
105 * bug fix: using obsoleted structure instances with TYPEP and
106 generic functions now signals a sensible error.
107 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
108 (reported by Kristoffer Kvello)
109 * bug fix: backtrace construction is now more careful when making
110 lisp-objects from pointers on the stack, to avoid creating bogus
111 objects that can be seen by the GC.
112 * bug fix: defaulting of values in contexts expecting more than 7
113 variables now works on x86-64. (reported by Christopher Laux)
114 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
116 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
117 as the property-list of a symbol.
118 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
119 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
120 situations requested, are once again file-compileable. (reported
123 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
124 * MOP improvement: support for user-defined subclasses of
125 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
126 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
127 the expansion of DEFMETHOD, is responsible for generating a form
128 which creates a list of specializers when evaluated. Additional
129 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
130 debugging and introspective support.
131 * minor incompatible change: the (unsupported) spinlock interface
132 has changed: free spinlock now has the value NIL, and a held spinlock
133 has the owning thread as its value.
134 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
135 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
137 * enhancement: name of a socket-stream is now "a socket" instead of
139 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
140 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
141 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
142 for generic functions. (Thanks to Tobias C. Rittweiler)
143 * optimization: bignum printing speed has been improved by 20-40%
144 (depending on the bignum size.)
145 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
147 * bug fix: the cache used by the CLOS to store precomputed effective
148 methods, slot offsets, and constant return values is now thread and
150 * bug fix: generic function dispatch function updating is now thread
151 and interrupt safe (in the sense that the known issues have been
153 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
154 * bug fix: interning EQL-specializers is now thread and interrupt safe.
155 * bug fix: asdf systems with dependencies to the SB-POSIX or
156 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
157 * bug fix: interrupt safety of applicable method computation has been
160 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
161 * new contrib: sb-cover, an experimental code coverage tool, is included
163 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
164 significantly faster.
165 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
166 produce more efficient code on x86-64 if the compiler can determine
167 that (+ INDEX <constant>) does not require a bounds check and FOO
168 has an element type at least 8 bits wide.
169 * enhancement: a new, experimental synchronous timeout facility is
170 provided. Refer to SB-SYS:WITH-DEADLINE for details.
171 * enhancement: when a symbol name conflict error arises, the
172 conflicting symbols are always printed with a package prefix.
173 (thanks to Kevin Reid)
174 * enhancement: stepping is now once again supported on the SPARC and
175 MIPS platforms. (It is also now more likely to work on CheneyGC
177 * enhancement: sb-sprof can now also track and report accurate call
179 * bug fixes: the treatment of non-standard subclasses of
180 SB-MOP:SPECIALIZER is more correct.
181 * incompatible change: PURIFY no longer copies the data from the
182 dynamic space into the static and read-only spaces on platforms
183 that use the generational garbage collector
184 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
186 * bug fix: GC race condition occasionally resulting in crashes with
187 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
189 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
191 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
192 a STYLE-WARNING to be signalled (regression from 1.0.4.)
193 * bug fix: an asynchronous interrupt could previously leave the
194 system running with GC inhibited.
195 * bug fix: a DECLARE form evaluated at top-level now causes an error
196 rather than silently (or verbosely) returning NIL.
197 * bug fix: trying to dissassemble functions compiled at high DEBUG could
198 result in a error being signalled due to source form lookup errors.
199 (reported by Peter Graves)
201 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
202 * incompatible change: removed writer methods for host-ent-name,
203 host-ent-addresses -- changing the values did not update the DNS
205 * minor incompatible change: changed experimental JOIN-THREAD interface
206 * documentation: the manual now lists reader and writer methods
207 in class slot documentation sections. (thanks to Richard M Kreuter)
208 * documentation: unwinding from asyncronous events has been
209 documented as unsafe.
210 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
211 in multithreaded application code.
212 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
213 platforms. (thanks to James Anderson for the optimization hint)
214 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
216 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
217 variants no longer cons.
218 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
219 their NOT- variants no longer cons.
220 * optimization: EQUAL hash tables no longer use SXHASH for objects
221 of all data types, but instead use an EQL hash for types for which
222 EQUAL is the same as EQL.
223 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
224 are significantly faster.
225 * optimization: new STRING-OUTPUT-STREAM implementation is much
226 faster for large outputs and conses 30% less on average.
227 * enhancement: XREF information is now collected for references made
228 to global variables using SYMBOL-VALUE and a constant argument.
229 * enhancement: SIGINT now causes a specific condition
230 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
231 * bug fix: FILE-POSITION used to signal an error in some cases where
232 ANSI requires it to return NIL.
233 * bug fix: ADJUST-ARRAY is now interrupt-safe.
234 * bug fix: adding and removing fd-handlers is now interrupt-safe.
235 * bug fix: inlined calls to C now ensure 16byte stack alignment on
237 * bug fix: bad type declaration in the CLOS implementation has
238 been fixed. (reported by James Anderson)
239 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
240 (thanks to Lutz Euler)
241 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
242 required by the CLHS. (reported by Samium Gromoff)
243 * bug fix: dead unbound variable references now signal an error.
244 (reported by Marco Monteiro)
245 * bug fix: / with an unused value was being deleted in safe code.
246 (thanks to Marco Monteiro and Kevin Reid)
247 * bug fix: number of characters that can be written onto a single
248 line in a file is unlimited.
249 * bug fix: some GC deadlocks caused by asynchronous interrupts have
250 been fixed by inhibiting interrupts for when GC is disbled.
251 * bug fix: some interrupt safety issues with GETHASH, (SETF
252 GETHASH), CLRHASH and REMHASH have been fixed.
253 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
254 specifier no longer causes infinite recursion.
255 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
256 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
257 * bug fix: an &environment argument with the correct variable information
258 is passed to macros that are expanded during byte compilation.
259 (reported by Samium Gromoff)
260 * bug fix: a logic error was causing the hash function for bignums to
261 have excessive amounts of collisions. (reported by Faré Rideau)
262 * bug fix: modifying the contents of an array could change the return
263 value of SXHASH on that array, which is only allowed for strings
264 and bit vectors. (bug introduced in 0.9.16)
265 * improvement: the x86-64/darwin port now passes all tests (except
266 for the debugger tests) but should still be considered
267 experimental until this is fixed.
268 * improvement: a style-warning is signaled for CASE (etc) clauses with
269 duplicate keys. (thanks to Kevin Reid)
270 * improvement: macos/x86 and macos/x86-64 now use mach exception
271 handlers for dealing with illegal instructions (for trapping and
272 error handling) and memory protection violations (for GC).
274 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
275 * new platform: experimental support for x86-64/darwin (MacOS).
276 * incompatible change: the thread-safe (on most platforms) getaddrinfo
277 and getnameinfo sockets functions are used instead of gethostbyaddr
278 and gethostbyname, on platforms where the newer functions are available.
279 As a result, the ALIASES field of HOST-ENT will always be NIL on these
281 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
282 don't print the startup banner, but behave as if --noinform was passed
283 as a command line argument. (thanks to Kevin Reid)
284 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
285 * optimization: code using alien values with undeclared types is much faster.
286 * optimization: the compiler is now able to open code SEARCH in more cases.
287 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
288 * bug fix: using standardized COMMON-LISP special variables as loop
289 variables no longer signals bogus package lock violations. (reported
291 * bug fix: declaring local loop variables to be of a range-limited type
292 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
293 (reported by Andras Simon)
294 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
295 bugs remain on x86-64.)
296 * bug fix: the #= and ## reader macros now interact reasonably with
297 funcallable instances.
298 * bug fix: type-checks for function arguments were compiled using the
299 compiler policy settings of the wrong lexical environment (in 1.0.2
301 * bug fix: SHADOW accepts characters as string designators, as required
302 by the spec (thanks to Eric Marsden)
303 * bug fix: fixed GC safety issues when foreign functions are called with
304 non-base strings as arguments
305 * bug fix: more consistent error output for fatal-compiler-warnings, like
307 * bug fix: functions with proper names were showing up as NIL in
308 backtraces (reported by Edi Weitz, regression in 1.0.2)
310 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
311 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
312 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
313 produce more efficient code on the x86 if the compiler can determine
314 that (+ INDEX <constant>) does not require a bounds check and FOO
315 has an element type at least 8 bits wide.
316 * bug fix: references to undefined variables are handled the same way
317 in toplevel forms as in the normal compiler
318 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
320 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
321 (thanks to Jon Buller)
322 * improvement: faster compilation times for complex functions
323 * improvement: added readlink support to SB-POSIX (thanks to Richard
326 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
327 * improvement: experimental support for mach exception handling on
328 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
330 * improvement: support for GBK external format.
331 (thanks to Chun Tian (binghe))
332 * improvement: the debugger now displays variables that have been closed
333 over, in code compiled with (DEBUG 2) or higher
334 * improvement: support for executable cores on NetBSD (thanks to
336 * new feature: added a RESTART-FRAME debugger command
337 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
338 be used to provide an implementation for FILE-POSITION on Gray streams
339 (thanks to Eric Marsden)
340 * new feature: add syslog, openlog and closelog support to SB-POSIX
341 (thanks to Richard Kreuter)
342 * optimization: the function call overhead in code compiled with
343 a high DEBUG optimization setting is significantly
344 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
345 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
346 (thanks to Tony Martinez)
347 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
348 evaluated once (reported by Kevin Reid)
349 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
350 works again (thanks to Kilian Sprotte)
351 * bug fix: an error is signaled for tagbodies with duplicate tags
352 (thanks to Stephen Wilson)
353 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
354 * bug fix: Win32 port can now handle foreign code unwinding Lisp
355 stack frames from alien callbacks.
356 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
357 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
358 65k pages (thanks to David Woodhouse)
359 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
361 changes in sbcl-1.0.1 relative to sbcl-1.0:
362 * new platform: FreeBSD/x86-64, including support for threading.
363 * new feature: the compiler stores cross-referencing information
364 abount function calls (who-calls), macroexpansion (who-macroexpands)
365 and special variables (who-binds, who-sets, who-references) for code
366 compiled with (< SPACE 3). This information is available through the
367 sb-introspect contrib.
368 * new feature: users may subclass SEQUENCE, and have instances of
369 these classes interoperate with standard Common Lisp functions if
370 a number of methods are defined. (This feature is experimental
371 and the interface subject to change based on feedback from SBCL
372 users and the general community)
373 * improvement: sb-sprof traces call stacks to an arbitrary depth on
374 x86/x86-64, rather than the previous fixed depth of 8
375 * improvement: another pthread back-end of mutex "pthread-futex".
376 * bug fix: non-ascii command-line arguments are processed correctly
377 (thanks to Yaroslav Kavenchuk)
378 * bug fix: non-required arguments were not passed correctly when a method
379 defined using DEFMETHOD was called from a mop-generated method using
380 CALL-NEXT-METHOD (reported by Pascal Costanza)
381 * bug fix: recursion is now permitted in accessors through
382 SLOT-UNBOUND. (reported by Pascal Costanza)
383 * bug fix: an error was signaled at startup if the HOME environment
384 variable was defined, but had an empty value (reported by Peter Van Eynde)
385 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
386 signal a PROGRAM-ERROR, not a BUG.
387 * bug fix: SB-POSIX:READDIR works when built with large file support.
388 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
390 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
391 for better automatic indentation support. (thanks to Matt Pillsbury)
392 * optimization: loading generic functions no longer takes O(n^2) time,
393 proportional to the amount of methods in the generic function
394 (reported by Todd Sabin and Jeremy Brown)
395 * optimization: the FIND and POSITION family of sequence functions
396 are significantly faster on arrays whose element types have been
398 * improvements to the Windows port:
399 ** Intermittent heap corruption problems have been fixed. (thanks
400 to Alastair Bridgewater)
401 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
403 ** Lisp is able to unwind foreign exception frames from alien
404 callbacks. (thanks to Alastair Bridgewater)
406 changes in sbcl-1.0 relative to sbcl-0.9.18:
407 * improvement: experimental support for threading on FreeBSD/x86.
408 (thanks to NIIMI Satoshi)
409 * improvement: runtime option --dynamic-space-size can be used
410 to set the size of the dynamic space reserved on startup.
411 * improvement: floating point modes in effect are now saved in
412 core, and restored on startup.
413 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
414 startup, not time since first call to GET-INTERNAL-REAL-TIME.
415 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
416 threads are not running after *SAVE-HOOKS* have run.
417 * improvement: writes to CLOS instance slots are type-checked in code
418 compiled with (SAFETY 3)
419 * improvement: floating-point exception handling on FreeBSD (thanks to
421 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
422 (thanks to Zach Beane)
423 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
425 * improvement: added support for the Shift-JIS external format.
426 (contributed by NIIMI Satoshi)
427 * improvement: callbacks are supported on Linux/PPC. (thanks to
429 * bug fix: compiler bug triggered by a (non-standard) VALUES
430 declaration in a LET* was fixed. (reported by Kaersten Poeck)
431 * bug fix: file compiler no longer confuses validated and already
432 dumped structurres. (reported by Kaersten Poeck)
433 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
434 fill-pointer signals a type-error as required. (thanks to
436 * bug fix: disassemly of funcallable instances works.
437 * bug fix: single stepping on PPC.
438 * bug fix: fix thread-safety problems in the type system (generally
439 manifesting as nonsensical errors like "STRING is a bad type specifier
440 for sequences" or "The value 1 is not of type FIXNUM")
441 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
442 (reported by Josip Gracin)
443 * bug fix: an error is signaled for attempts to displace arrays with
444 incompatible element types (thanks to Mario Mommer)
445 * bug fix: more correct handling of wide characters in debug info
446 (bug reported by Attila Lendvai and fixed by Juho Snellman)
447 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
448 and don't cause extra consing
449 * optimization: MAP and MAP-INTO are significantly faster on vectors
450 whose elements types have been declared.
451 * Improvements to SB-SPROF:
452 ** Support for allocation profiling
453 ** Reduced profiling overhead, especially for long profiling runs
454 * Improvements to the Windows port:
455 ** floating point exceptions are now reported correctly.
456 ** stack exhaustion detection works partially.
457 ** more accurate GET-INTERNAL-REAL-TIME.
458 ** better RUN-PROGRAM behaviour in case of errors in the child
460 ** PROBE-FILE now simplifies pathnames correctly.
461 ** DIRECTORY now works correctly with :WILD-INFERIORS.
462 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
463 1st 1970 (although time timezone data will be unreliable).
464 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
466 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
467 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
468 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
470 * bug fix: two potential GC deadlocks affecting threaded builds.
471 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
472 returns T (reported by Anton Kazennikov)
473 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
474 with non-variable places
475 * bug fix: remove a race condition in the setting of
476 funcallable-instance functions, this should make threaded CLOS
477 code more stable against memory faults.
478 * bug fix: corruption of specials when unbinding is interrupted by an
479 asynchronous unwind (reported by Hannu Koivisto)
480 * improvement: the debugger will now also display local variables that
481 are only used once, for code compiled with a DEBUG optimization quality
484 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
485 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
486 * incompatible change: External-format support for FFI calls. The
487 SB-ALIEN:C-STRING no longer implies an ASCII
488 external-format. Instead the string is subject to external-format
489 conversion. Additionally return values of type SB-ALIEN:C-STRING
490 are by default of type (SIMPLE-ARRAY CHARACTER), not
491 SIMPLE-BASE-STRING. For an alien type that behaves like the old
492 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
493 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
494 * incompatible change: SB-EXT package no longer contains the
495 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
496 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
498 * incompatible change: the single-stepper is no longer available
499 on Alpha, Mips and Sparc platforms.
500 * minor incompatible change: the direct superclasses of
501 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
502 not (STANDARD-OBJECT FUNCTION). This makes the
503 class-precedence-lists of GENERIC-FUNCTION and
504 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
506 * minor incompatible change: the default stream external format on
507 non-unicode SBCL is no longer detected from the locale, but is always
509 * new feature: Add a version of evaluator that uses an interpreter instead
510 of the compiler. EVAL still uses the compiler by default, to switch it
511 to use the interpreter, set the value of the variable
512 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
513 * minor incompatible change: the single-stepper REPL has been merged
514 with the normal debugger (see the "Stepping" heading of the debugger help
515 for more details). The debugger command STEP will no longer switch
516 to the single-stepper REPL.
517 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
518 for a type now works.
519 * bug fix: loading of default sysinit file works. (thanks to Leonid
521 * bug fix: better detection of circularities in the file-compiler.
522 (reported by Marco Monteiro)
523 * bug fix: the CL pathname functions now work with files that have
524 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
525 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
526 systems with Unix98 pty semantics.
527 * bug fix: ASDF-INSTALL will now work with bsd tar.
528 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
530 * bug fix: timers expiring in dead threads no longer cause a
531 type-error (reported by Paul "Nonny Mouse").
532 * bug fix: thanks to more lightweight single-stepper instrumentation,
533 code compiled with (DEBUG 3) will compile and execute significantly faster,
534 and will have more accurate type-inferencing than before
535 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
536 whose bindings are modified
537 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
538 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
539 ** replace dummy implementations of CL:MACHINE-INSTANCE and
540 CL:SOFTWARE-VERSION with proper version
542 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
543 * feature: implemented the READER-METHOD-CLASS and
544 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
545 as specified by AMOP.
546 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
548 * optimization: faster LOGCOUNT implementation on x86 and x86-64
549 (thanks to Lutz Euler)
550 * optimization: hashing of general arrays and vectors has been
551 improved. (reported by Any Fingerhut)
552 * enhancement: SB-INTROSPECT is now able to find definitions of
553 profiled functions. (thanks to Troels Henriksen)
554 * enhancement: compiler-macro expansion applies now to FUNCALL forms
555 as well, allowing compiler-macros for SETF-functions to expand.
556 * enhancement: step-instrumentation no longer wraps known
557 single-value functions with multiple-value context, allowing
558 better type inference.
559 * fixed bug #337: use of MAKE-METHOD in method combination now works
560 even in the presence of user-defined method classes. (reported by
561 Bruno Haible and Pascal Costanza)
562 * fixed bug #339(c): if there are applicable methods not part of any
563 long-form method-combination group, call INVALID-METHOD-ERROR.
564 (reported by Bruno Haible)
565 * fixed bug #361: the :FUNCTION initarg in the protocol for
566 initialization of methods can now be used to override
567 internally-produced optimized functions. (reported by Bruno
569 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
570 system-provided lambda expression no longer cause warnings about
571 unbound #:|pv-table| symbols.
572 * bug fix: improved the handling of type declarations and the
573 detection of violations for keyword arguments with non-constant
575 * bug fix: potentially erronous calls to PATHNAME and
576 MERGE-PATHNAMES were being flushed in some cases.
577 (reported by Richard Kreuter)
578 * bug fix: compiled calls to TYPEP were mishandling obsolete
579 instances. (reported by James Bielman and Attila Lendvai)
580 * bug fix: format strings with ~> without matching ~< no longer
581 trigger an AVER, but signal an understandable error instead.
582 (reported by Antonio Martinez)
583 * bug fix: specifying an output-file without a pathname-type for
584 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
585 of input-file instead of "fasl". (reported by Robert Dodier)
586 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
587 (reported by James Y Knight).
588 * bug fix: compiler-macros-function did not consider the environment
589 argument for shadowing by local functions.
590 * bug fix: compiler-macros expansion was inhibited by local INLINE
592 * bug fix: inline expansions of known functions were subject to
593 step-instrumentation in high debug policies, leading to problems
595 * bug fix: compiler failed to differentiate between different CONS
597 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
598 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
599 * bug fix: #368: incorrect use of expressed vs. upgraded array
601 * bug fix: #306a: more precise unions of array types.
602 * thread-safety improvements:
603 ** CONDITION-WAIT could return early on Linux, if the thread was
604 interrupted and subsequently continued with SIGCONT.
605 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
607 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
608 * added support for the ucs-2 external format. (contributed by Ivan
610 * minor incompatible change: pretty printing of objects of type
611 (cons symbol) is, in the default pprint-dispatch-table, now
612 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
614 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
615 called later in a class's lifetime, possibly as late as when the
616 first instance of the class is created. Previously,
617 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
618 class became finalizeable.
619 * fixed bug: FILE-POSITION sometimes returned inconsistent results
620 for multibyte external-format streams. (thanks to "vbzoli")
621 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
622 with :ALLOCATION :CLASS inherited from superclasses of the
624 * fixed bug: anonymous classes can now be created using the :NAME
625 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
626 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
627 * fixed bug: core-files saved with :EXECUTABLE T can again be
628 executed when SBCL_HOME isn't set. (reported by James Knight)
629 * fixed bug: toplevel LOCALLY forms with declarations could
630 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
631 * fixed bug: printing from several different threads using different
632 values of *print-case* could cause invalid output, due to
633 some internal special variables of the printer not being bound
634 thread-locally (reported by Max Mikhanosha)
635 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
636 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
637 (reported by Pascal Costanza)
638 * fixed bug: FINALIZE-INHERITANCE is called from
639 REINITIALIZE-INSTANCE on classes when the class has previously
640 been finalized, as required by AMOP.
641 * minor code generation optimizations:
642 ** better register allocation in CLOS dispatching functions
643 ** overflow detection when coercing signed bytes to fixnums on x86-64
644 ** is now implemented with one IMUL instruction instead of three shifts
645 ** more efficient bit-vector access on x86 and x86-64
646 ** more efficient access to raw structure slots on x86 and x86-64
647 * fixed some bugs revealed by Paul Dietz' test suite:
648 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
651 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
652 * feature: thread support on Solaris/x86, and experimental thread support
654 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
656 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
657 default method for SLOT-UNBOUND.
658 * minor incompatible change: prevent the user from specializing the
659 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
660 somewhat counter to the intent of the protocol, I (CSR) think, and
661 additionally it just doesn't work in SBCL as currently
662 implemented, thanks to optimizations (that are always valid for
663 the other three SLOT-VALUEish functions, but not for the setter).
664 * minor incompatible change: the :SB-LDB feature is now enabled by
665 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
666 the low-level debugger.
667 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
668 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
669 down to an SB-SYS:FD-STREAM, allowing direct communication in
670 more cases. Temporary files and pipes are still used as a fallback
672 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
673 on the non-reentrant C functions localtime(3) and gmtime(3).
674 * bug fix: global optimization policy was not visible in LOCALLY and
676 * bug fix: class objects can be used as specializers in methods.
677 (reported by Pascal Costanza)
678 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
679 with a type error. (reported by blitz_ on #lisp)
680 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
681 failed with a type error. (reported by Pascal Bourguignon)
682 * bug fix: merging pathnames against defaults with :DIRECTORY
683 starting with '(:RELATIVE :BACK) should preserve the :BACK.
684 (reported by James Y Knight)
685 * bug fix: saving large (>2GB) cores on x86-64 now works
686 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
687 constant index and a (simple-array (signed-byte 32)) array
688 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
689 error (patch by Robert J. Macomber)
690 * feature: TIME now displays the amount of run-time spent in GC
691 * bug fix: The debugger now does a better job of respecting
692 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
693 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
694 workaround for bug 403.)
695 * bug fix: floating point exception handling now works on FreeBSD
696 (thanks to NIIMI Satoshi)
697 * fixed some bugs revealed by Paul Dietz' test suite:
698 ** MISC.641: LET-conversion were not supposed to work in late
700 * improvements to the Win32/x86 port:
701 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
702 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
704 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
705 * new feature: source path information is generated for macro-expansion
706 errors for use in IDE's like Slime (thanks to Helmut Eller)
707 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
708 * bug fix: compiler does not loop forever on an invalid type in
710 * improvement: compilation of most CLOS applications is significantly
712 * optimization: added a limited bytecode compiler for simple toplevel
713 forms, speeding up compilation and FASL loading
714 * bug fix: the statistical profiler now properly distinguishes anonymous
717 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
718 * minor incompatible change: in sbcl-0.9.11 (but not earlier
719 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
720 into subclasses of the finalized class. Now user calls to
721 FINALIZE-INHERITANCE finalize just the one class, and calls by the
722 system return before any subclasses are finalized.
723 * minor incompatible change: The reader no longer ignores errors
724 regarding non-existent packages in #+ and #- feature tests.
725 * new feature: command line options --no-sysinit, --no-userinit to
726 inhibit loading the corresponding init files
727 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
728 for sending data through UDP sockets (thanks to François-René Rideau)
729 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
730 error is signalled instead (thanks to François-René Rideau)
731 * minor incompatible change: Error signalling behaviour of lexical
732 operations violating package locks has changed slightly. Refer to
733 documentation on package locks for details.
734 * enhancement: EVAL can process IF-expressions without resorting to the
736 * bug fix: Some operations involving SIN, COS, or TAN weren't being
737 constant-folded properly due to missing out-of-line functions.
738 (reported by Mika Pihlajamäki)
739 * bug fix: LISTEN sometimes returned T even in cases where no data was
740 immediately available from the stream
741 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
742 were reversed. (reported by Levente Mészáros)
743 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
744 when given a vector displaced to an adjustable array. (reported
746 * bug fix: derivation of float boundaries from numbers outside the
747 appropriate float range (reported by John Wiseman)
748 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
750 * bug fix: precomputing cache entries for generic functions with
751 some subclasses of specializers as yet invalid does not attempt to
752 fill a cache line with a negative offset. (reported by Levente
754 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
755 allow condition class objects as arguments to DOCUMENTATION and
756 (SETF DOCUMENTATION); only find and set documentation for
757 structure names for the STRUCTURE doc-type. (suggested by Gary
759 * improvements to the Win32/x86 port:
760 ** (user-homedir-pathname) and default initialization file
761 locations now know about the user's "Documents and Settings"
762 directory (thanks to Yaroslav Kavenchuk)
763 ** run-program is implemented (thanks to Mike Thomas)
764 ** sockets support (thanks to Timothy Ritchey)
765 ** better backtrace support (thanks to Alastair Bridgewater)
766 ** sb-grovel supported
767 ** asdf-install and sb-posix work somewhat
768 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
769 * improvements to the Solaris/x86 port:
770 ** works on Solaris 11/Solaris Express
771 ** floating-point exception handling support
772 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
773 * fixed some bugs revealed by Paul Dietz' test suite:
774 ** REMOVE-METHOD returns its generic function argument even when
775 the method is not one of the generic functions' methods.
776 ** objects declared as MEMBER types can be admissible arguments to
778 ** printing characters should simply be printed by the FORMAT ~:C
780 ** compiler failure when compiling functions with hairy constant
781 defaults for optional parameters.
782 ** compiler produces wrong code when MAYBE-INLINE-expanding a
783 function, which is already optimized.
785 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
786 * new platform: experimental support for SBCL x86/Darwin, including
787 MacOS X 10.4.5 on Intel.
788 * new feature: Unicode character names are now known to the system
789 (through CHAR-NAME and NAME-CHAR).
790 * new feature: the filesystem location of SBCL's core file is
791 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
792 * minor incompatible change: the contrib modules SB-POSIX and
793 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
794 this change is to make it easier to distribute
795 location-independent binaries.
796 * bug fix: as implied by AMOP, standardized classes no longer have
797 slots named by external symbols of public packages. (reported by
799 * optimization: calling functions via constant symbols -- as in
800 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
801 via the function object as in (FUNCALL #'FOO).
802 * optimization: on x86 and x86-64, the calling convention for the
803 case of dealing with an unknown number of values has been altered
804 to be friendlier to the prediction heuristics implemented,
805 particularly with reference to CALL and RET pairing. (thanks to
806 Alastair Bridgewater)
807 * optimization: on x86, the code for access to thread-local symbol
808 values and binding/unbinding of thread-local symbols is smaller.
809 (thanks to Alastair Bridgewater)
810 * enhancement: CONSTANTP is now able to determine constantness of
811 more complex forms, including calls to constant-foldable standardized
812 functions and some special forms beyond QUOTE.
813 * fixed bug: occasional GC crashes on Solaris/x86
814 * optimization: x86-64 supports stack allocation of results of simple
815 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
816 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
817 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
818 (thanks to James Knight)
819 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
820 by the standard (thanks to Stephen Compall)
822 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
823 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
824 be used for bundling the runtime and the core file into one
825 executable binary. This feature is not currently supported on all SBCL
826 platforms. (thanks to James Bielman and NIIMI Satoshi)
827 * new feature: a generational or ephemeral garbage collector is now
828 the default on the PowerPC platform (both Linux and Darwin). The
829 old Cheney (stop and copy) collector is a build-time option.
830 (thanks to Cyrus Harmon, after Raymond Toy)
831 * minor incompatible change: the method by which SBCL finds its
832 contributed modules has changed; it no longer relies on symbolic
833 links from an $SBCL_HOME/systems directory, but searches directly
834 in the subdirectories of $SBCL_HOME.
835 * enhancement: the dynamic heap size on the Linux/PPC platform is
836 markedly larger, even using the older Cheney garbage collector.
837 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
838 certain complicated string types. (reported by Gary King)
839 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
840 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
841 * fixed bug: DOCUMENTATION on structures no longer signals an
842 error if no structure type of the right name exists. (reported by
844 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
845 applicable non-standard primary or :AROUND method on
846 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
847 no longer cause unbound variable errors. (reported by Kevin Reid)
848 * fixed bug: metacircle resolution in cases where methods have slots
849 added before the slots from STANDARD-METHOD. (reported by Jean
851 * fixed bug: the Power PC assembly code for calling into lisp
852 sometimes computed its return address wrongly (depending on the
853 15th bit of the address, fixed at link time). This used to
854 manifest itself in a segmentation violation while building PCL.
855 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
856 many others over the years)
857 * fixed bug: ROOM no longer reports silly numbers for stack usage if
858 the stack pointer (treated as a signed integer) is negative.
859 (thanks to Peter van Eynde)
861 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
862 * new platform: experimental support for the Windows operating
863 system has been added. (thanks to Alastair Bridgewater)
864 * fixed several bugs in and robustified the PPC FFI (including
865 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
866 * bug fix: allow non-simple string symbol names (reported by Paul
868 * bug fix: interrupt handling on NetBSD (thanks to Richard M
870 * bug fix: saving a core corrupted callbacks on x86/x86-64
871 * bug fix: closed a loophole in metacircularity detection and
872 grounding in the PCL implementation of CLOS.
873 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
874 (reported by tomppa on #lisp)
875 * bug fix: clear the direction flag on Lisp -> C transitions, as
876 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
877 (reported by Andrej Grozin and Hendrik Maryns)
878 * optimization: major improvements to GC efficiency on GENCGC platforms
879 * optimization: faster implementation of EQUAL
880 * optimization: emit more efficient opcodes for some common
881 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
882 * optimization: several other minor code-generation improvements on x86-64
883 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
885 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
886 * minor incompatible change: (SETF CLASS-NAME) and (SETF
887 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
888 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
889 by ANSI as a generic function, and for consistency (SETF
890 GENERIC-FUNCTION-NAME) is treated likewise.
891 * fixed bug #233.b: make constraint propagation notice when a variable
892 value is changed after it is referenced but before it is used
893 * fixed bug #296: no more arbitrary behaviour from filenames with
894 odd characters as --load arguments. (reported by Adam Warner)
895 * fixed bug #390: :CHARACTER-SET pathname components now work as
896 expected. (reported by Tim Daly Jr)
897 * fixed bug #391: complicated :TYPE intersections in slot
898 definitions no longer cause an error in PCL internals.
899 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
900 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
901 returning the number of octets which would be written to the
902 file-stream. (thanks to Robert J. Macomber)
903 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
904 arguments to the internal stream functions. (thanks to David
906 * bug fix: stack exhaustion is now handled on the x86/SunOS
907 platform. (thanks to Daisuke Homma)
908 * bug fix: more accurate ROOM results on GENCGC platforms
909 * optimization: improved type inference for arithmetic-for
910 index variables in LOOP
911 * optimization: faster floating-point SQRT on x86-64
912 * fixed some bugs revealed by Paul Dietz' test suite:
913 ** DOCUMENTATION returns NIL instead of "" for method combinations
914 that don't have a docstring
916 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
917 * minor incompatible change: (SETF CLASS-NAME) and (SETF
918 GENERIC-FUNCTION-NAME) are no longer generic functions, and
919 therefore are not customizeable by user code (as seems to be at
920 least permitted and maybe required by AMOP). As a consolation,
921 however, the SBCL implementation of these functions now calls
922 REINITIALIZE-INSTANCE as specified by AMOP.
923 * bug fix: slot-definition documentation is propagated as per ANSI
924 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
925 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
926 argument, as do eslotd initialization methods. (from Pascal
927 Costanza's "Closer" project)
928 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
929 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
931 * bug fix: it is now possible to have more than one subclass of
932 STANDARD-GENERIC-FUNCTION without causing stack overflow.
933 (reported by Bruno Haible, Pascal Costanza and others)
934 * bug fix: the dependent update protocol now works for generic
935 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
937 * bug fix: condition-class instances corresponding to
938 DEFINE-CONDITION forms are now created eagerly. (reported by
939 Kalle Olavi Niemitalo on comp.lang.lisp)
940 * bug fix: floating point printing is more accurate in some
941 circumstances. (thanks to Simon Alexander)
942 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
943 merged with *DEFAULT-PATHNAME-DEFAULTS*.
944 * bug fix: callbacks on OS X now preserve stack-alignment.
945 (thanks to Cyrus Harmon)
946 * enhancement: the x86-64 disassembler is much better at
947 disassembling SSE instructions. (thanks to Lutz Euler)
948 * enhancement: CHECK-TYPE on a variable now allows type-inference
949 (thanks to Vincent Arkesteijn)
950 * optimization: improved performance of EUC-JP external format.
951 (thanks to NIIMI Satoshi)
952 * optimization: performance improvements to IO on file streams of
953 :ELEMENT-TYPE CHARACTER
954 * optimization: much faster memory allocation on x86-64
955 * optimization: faster garbage collections (latency and throughput)
957 * optimization: faster bignum subtraction and division on x86 and x86-64
958 (thanks to Lutz Euler)
959 * optimization: more accurate type-inference for loops that use a
960 floating point index variable or a negative step.
962 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
963 * bug fix: add a workaround to SBCL looping infinitely at startup on
964 Linux kernels with apparently buggy implementations of personality().
965 (thanks to Svein Ove Aas)
966 * bug fix: Unicode symbols are correctly printed in LDB backtraces
967 (thanks to David Lichteblau)
968 * bug fix: local bindings shadow global symbol macros for
969 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
970 * optimization: non-open coded uses of numeric comparison operators
971 (e.g. >) no longer cons when called with more than one parameter
972 on platforms supporting dynamic-extent allocation.
973 * enhancement: saving cores with foreign code loaded is now supported
974 on MIPS/Linux in addition to the previously supported platforms.
975 * bug fix: FIND-RESTART now tests for activity, not applicability when given
976 a restart object as identifier. (reported by Helmut Eller for CMUCL)
977 * bug fix: division by zero in sb-sprof when no samples were collected
978 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
979 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
980 (thanks to Alexander Kjeldaas)
981 * new feature: ignoring the return values of destructive functions
982 like NREVERSE signals a compile-time style-warning.
983 (thanks to Kevin Reid)
984 * new feature: the alignment of alien structure fields can be
985 explicitly specified. (thanks to Cyrus Harmon)
986 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
987 and dump core on SIGQUIT
989 ** incompatible change: threads do not inherit values of specials
990 from their parents (see manual)
991 ** bug fix: threads stacks belonging to dead threads are freed by the
992 next exiting thread, no need to gc to collect thread stacks anymore
993 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
994 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
995 ** bug fix: dynamic variable and thread start related gc lossage
996 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
998 * fixed some bugs revealed by Paul Dietz' test suite:
999 ** SUBTYPEP is slightly more accurate on heinously complicated
1000 CONS types where some of the members have uncertain (in the
1001 NIL, NIL sense) type relationships to each other.
1003 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1004 no longer purified unless :PURIFY T is explicitly specified.
1005 ** Non-purified cores are significantly smaller than before
1007 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1008 * new feature: timers based on Zach Beane's excellent timer package
1009 * added support for the following external formats: koi8-u,
1010 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1011 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1012 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1013 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1014 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1015 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1016 * incompatible change: a threaded SBCL will no longer revert to
1017 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1018 * bug fix: interrupts are disabled until startup is complete; no
1019 more sigsegvs when receiving a signal to soon
1020 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1022 * bug fix: add a workaround for the memory randomization features in
1023 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1024 workaround will only be in effect on systems with the proc filesystem
1026 * bug fix: printing objects of type HASH-TABLE signals a
1027 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1029 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1030 have been running for over 50 days. (reported by Gilbert Baumann)
1031 * bug fix: the logic for getting names of functions gets less
1032 confused when confronded with alternate-metaclass
1033 funcallable-instances. (reported by Cyrus Harmon)
1034 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1035 consistent, even on internal alternate-metaclass objects.
1036 * bug fix: flush closure information collected by physical
1037 environment analysis prepass before the main pass. (bug reported
1039 * bug fix: compiler pack phase does not modify a hashtable, which is
1040 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1041 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1042 may be expanded. (reported by Alexander <malishev> on c.l.l)
1043 * bug fix: no more style-warnings when compiling code using LAST
1044 under high-SPEED low-SPACE settings. (reported by David Wallin
1046 * fixed some incompatibilities between SBCL's MOP and the MOP
1048 ** the METAOBJECT class is now implemented;
1049 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1050 STANDARD-OBJECT, as required;
1051 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1052 now compatible (as required by VALIDATE-SUPERCLASS); there
1053 remains a consistency requirement over the presence or absence
1054 of the FUNCTION class in the superclass list of finalized
1055 classes; see the manual for more details;
1056 ** the :AROUND method for COMPUTE-SLOTS on
1057 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1058 requested slot ordering.
1060 ** bug fix: parent thread now can be gc'ed even with a live
1062 ** bug fix: binding a special with PROGV to no value is not
1063 the same as that symbol not having been bound (thanks to
1065 ** bug fix: binding specials is thread safe (thanks to
1067 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1068 and SB-SPROF do not die with 'no handler for signal XX in
1069 interrupt_handle_now(..)' anymore
1070 ** bug fix: WITH-TIMEOUT works with multiple running threads
1071 * fixed some bugs revealed by Paul Dietz' test suite:
1072 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1073 the :method-class keyword argument.
1075 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1076 * new port: the Solaris operating system on x86 processors is now
1077 mostly supported, though some rough edges in the environment
1078 remain. (thanks to Daisuke Homma)
1079 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1080 space for its heap. (thanks to Thiemo Seufer)
1081 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1082 closures. (thanks to Thiemo Seufer)
1083 * minor incompatible change: eof selects abort in the debugger.
1084 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1085 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1087 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1088 deprecated, and will go away in a future revision of SBCL.
1089 * minor incompatible change: GC-ON and GC-OFF are no longer
1090 implemented with a counter, it does not matter how many times gc
1091 is switched on or off
1092 * bug fix: discriminating functions for generic function classes
1093 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1094 make invalid assumptions about method precedence order. (reported
1096 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1097 platforms. (thanks to Thiemo Seufer)
1098 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1099 * bug fix: correct pathname printing with printer escaping is on.
1100 (thanks to Kevin Reid)
1101 * bug fix: complex VOP definitions in "user-space" no longer trigger
1102 package locks. (reported by Zach Beane)
1103 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1104 for funcallable-instance functions, and is no different from
1106 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1107 characters in the current readtable. (reported by Nicholas Neuss)
1108 * bug fix: SUBTYPEP on various CONS types returns more a more
1109 accurate acknowledgment of its certainty.
1110 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1111 lists in some cases. This partially fixes bug 384.
1112 * flush all standard streams before prompting in the REPL and the
1114 * bug fix: signal handling and triggering gc do not conflict
1115 directly anymore, in particular a high frequency sb-sprof does
1116 not prevent gc from running
1117 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1118 approximation for timezone and DST information between the
1119 universal time epoch and the smallest negative 32-bit time_t.
1120 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1121 year 1899 so long as the encoded time is non-negative.
1122 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1123 the correct answer for non-constant rotate amounts when performing
1124 an inline 32-bit rotation.
1126 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1127 there is only one thread in the session
1128 ** bug fix: memory leak for streams created in one thread and
1129 written to in another
1130 ** bug fix: lockup when compiled with gcc4
1131 ** bug fix: race that allows the gc to be triggered when gc is
1133 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1134 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1136 ** bug fix: finished threads are gc'ed properly
1137 * fixed some bugs revealed by Paul Dietz' test suite:
1138 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1139 has arguments with a different set of applicable methods from
1140 the orignal arguments.
1141 ** The type error thrown by MAP now has a correct expected-type
1143 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1144 name a compiled function.
1145 ** Three MISC tests where a large bignum was improperly coerced to
1146 a float that couldn't represent that bignum during type
1147 derivation were fixed.
1148 ** SUBTYPEP can now handle types involving the negation of a
1149 list-form FUNCTION type.
1150 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1151 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1152 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1154 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1155 * New feature: Experimental support for bivalent streams: streams
1156 opened with :element-type :default now allow character and binary
1157 (unsigned-byte 8) I/O
1158 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1159 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1161 * bug fix: no more highly sporadic "couldn't check whether ~S is
1162 readable" when reading a stream and an interrupt hits in the middle
1163 of a select system call
1164 * compiler better recognizes complex arrays (reported by Eduardo
1166 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1167 for variables on SBCL built with linkage-tables. (reported by Luis
1169 * various error reporting improvements.
1170 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1171 (Thanks to Thiemo Seufer)
1172 * bug fix: nested reader invokations work correctly
1173 * bug fix: it is possible to have simultaneous references to foreign
1174 code and foreign data with the same name.
1176 ** added x86-64 support
1177 ** incompatible change: the threading api now works with thread
1178 objects instead of thread ids
1179 ** bug fix: threads are protected from signals and interruption when
1180 starting up or going down
1181 ** bug fix: a race where an exiting thread could lose its stack to gc
1182 ** bug fix: don't halt on infinite error in threads if possible
1183 ** fixed numerous gc deadlocks introduced in the pthread merge
1184 ** bug fix: fixed thread safety issues in read and print
1185 ** bug fix: debugger doesn't hang on session lock if interrupted at
1186 an inappropriate moment
1187 ** bug fix: run-program is now thread safe(r)
1188 ** bug fix: inner with-recursive-lock no longer releases the mutex
1189 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1190 * fixed some bugs revealed by Paul Dietz' test suite:
1191 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1192 EXPECTED-TYPE slots filled.
1193 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1195 ** MAKE-CONDITION accepts classes as type-designators.
1196 ** COMPILE may never return NIL.
1197 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1198 range before calling Unix time functions
1200 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1201 * numerous signal handling fixes to increase stability
1202 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1203 * minor incompatible change: we now correctly canonize default
1204 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1205 as per the MOP, rather than the historical (INITARG INITFUNCTION
1206 INITFORM). (reported by Bruno Haible)
1207 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1208 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1209 for more information.
1210 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1211 pathname is a directory pathname.
1212 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1213 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1215 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1216 :METACLASS argument in addition to a class name. (reported by
1217 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1218 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1219 (reported by Bruno Haible; patch from Gerd Moellmann)
1220 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1222 * bug fix: more cleanups to the floating point exception handling on
1223 x86-64 (thanks to James Knight)
1224 * bug fix: the compiler does not try to constant fold calls of
1225 COERCE and BIT-* functions when they return freshly constructed
1226 objects. (reported by Nikodemus Siivola and Paul Dietz)
1227 * optimization: improved the MIPS versions of generic (in the
1228 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1229 * optimization: direct conversion of (unsigned-byte 32) to floats on
1230 the PowerPC platform.
1231 * optimization: structure instances with raw slots now use less
1232 memory, and probably show better memory locality. (thanks to
1234 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1235 are treated as requests for stack allocation on the x86-64,
1236 Alpha-32, PPC and SPARC platforms.
1237 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1238 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1239 (thanks to Tony Martinez)
1241 ** gcing a dead thread can no longer lead to lockups
1242 ** threads block signals until they are set up properly
1243 ** errno is no longer shared by threads
1244 ** interrupt-thread restores the eflags register on x86
1245 ** fixed some lockups due to gc/thread interaction
1246 * fixed some bugs revealed by Paul Dietz' test suite:
1247 ** invalid dotted lists no longer raise a read error when
1248 *READ-SUPPRESS* is T
1249 ** signal an error if a symbol that names a declaration is used
1250 as the name of a type, or vice versa
1251 ** allow using the (declare (typespec var*)) abbreviation for
1252 (declare (type typespec var*)) with all type specifiers
1253 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1254 ** always return NIL from PPRINT-POP when OBJECT is NIL
1255 ** don't signal errors when pretty-printing malformed LABELS,
1256 FLET or MACROLET forms
1257 ** declarations in a DOLIST body are also in scope for the
1259 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1261 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1264 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1265 * fixed cross-compiler leakages that prevented building a 32-bit
1266 target with a 64-bit host compiler.
1267 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1268 opened with :IF-EXISTS :OVERWRITE.
1269 * fixed bug 281: error for an invalid qualifier in a short-form method
1270 combination method is not signalled until the faulty method is called.
1271 * bug fix: iteration variable type inferrer failed to deal with open
1272 intervals. (reported by Alan Shields)
1273 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1274 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1275 or more runtime options were provided to the sbcl binary.
1276 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1277 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1278 * optimization: there's now a fast-path for fixnum arguments in the
1279 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1280 * optimization: the code generated on x86-64 is more compact thanks
1281 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1282 * bug fix: floating-point exception handling now partly works on
1283 x86-64. (Thanks to James Knight)
1284 * improvement to the MIPS backend from Thiemo Seufer: C-style
1285 64-bit long long arguments and return values to alien functions
1287 * fixed some bugs revealed by Paul Dietz' test suite:
1288 ** the type-error signalled from WARN has a filled-in DATUM slot.
1289 ** the type-error required when a stream is not associated with
1290 a file has the stream as its datum.
1291 ** type-errors on single-floats on x86-64 no longer have
1292 :INVALID-OBJECT as the datum
1293 ** the type-errors signalled for invalid function names now have
1294 a correct expected type
1295 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1296 for structures defined with a :TYPE. Documentation strings for
1297 typed structures are no longer immediately discarded
1298 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1299 broadcast streams and synonym streams. FILE-LENGTH now also works
1300 on broadcast streams.
1302 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1303 * incompatible change: the --noprogrammer option, deprecated since
1304 version 0.7.5, has been removed. Please use the equivalent
1305 --disable-debugger option instead.
1306 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1308 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1309 inoperational for a while now) has been completely removed.
1310 * null lexical environments are now printed as #<NULL-LEXENV>,
1311 significantly reducing the amount of clutter in typical backtraces.
1312 * documentation on weak pointers, finalization, and after GC hooks
1313 has been added to the manual.
1314 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1315 as other specialized array types, is much faster. SUBSEQ and
1316 COPY-SEQ on such arrays have also been sped up.
1317 * optimization: EQL is now more efficient when at least other argument
1318 is known to be of type (OR FIXNUM (NOT NUMBER)).
1319 * fixed bug: the runtime is now less vulnerable to changes in the
1320 size of the SBCL object on OS X, and virtual memory is reserved for
1321 all spaces that need to be at a fixed address.
1322 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1323 * fixed bug: finalizers and after GC hooks that cause consing are now
1325 * fixed bug: compiler error messages and summaries are now printed to
1326 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1327 * fixed inference of the upper bound of an iteration variable.
1328 (reported by Rajat Datta).
1329 * fixed bug 211e: calling local functions with duplicated constant
1330 keyword argument no longer causes a bogus style warning about an
1332 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1333 ftype declaration to be disregarded. (reported by Dave Roberts)
1334 * fixed bug 373: caused by erronous compilation of references to alien
1335 variables in the runtime on ppc/darwin.
1336 * fixed bug 376: CONJUGATE type deriver.
1337 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1338 (reported by Baughn on #lisp)
1339 * a cross-compiler bug on non-x86 platforms has been identified and
1340 fixed. (thanks to Bruno Haible)
1341 * improvements to the MIPS runtime code for increased stability.
1342 (thanks to Thiemo Seufer)
1343 * increased the maximimum compact environment size to allow
1344 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1345 * partial workaround for bug 135: don't name the function we're
1346 calling for hairy cases of EVAL, so as not to accumulate one environment
1347 entry per call to EVAL. (reported by Kevin Reid)
1348 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1349 * optimization: type testing for non-vector arrays should be faster.
1350 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1352 * bug fix: setting 31st element of a bit vector to zero did not work
1354 * bug fix: redefining a class definition which failed due to a
1355 previous accessor / function clash now works (but see BUGS entry
1356 #380 for more problems in this area). (thanks to Zach Beane)
1357 * the long form of DEFINE-METHOD-COMBINATION disables method group
1358 checking when given a single method group with pattern *. (thanks
1360 * on x86 compiler supports stack allocation of results of simple
1361 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1362 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1363 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1364 * fixed some bugs related to Unicode integration:
1365 ** the restarts for recovering from input and output encoding
1366 errors only appear when there is in fact such an error to
1368 * fixed some bugs revealed by Paul Dietz' test suite:
1369 ** MISC.549 and similar: late transformation of unsafe type
1370 assertions into derived types caused unexpected code
1372 ** SCALE-FLOAT type deriver is less wrong.
1373 ** type derivers for EXP, LOG and similar functions compute result
1374 types for complex arguments better.
1375 ** (MISC.563) CONJUGATE type deriver works for very restricted
1377 ** out-of-line type testers for character strings are available.
1378 ** EQUAL compiler transform understands specialness of objects
1380 ** accessing double-floats stored on the stack now works on x86-64.
1381 ** debugger internals could sometimes create invalid lispobjs,
1382 resulting in GC crashes.
1383 ** MISC.548: type check weakening can convert required type into
1385 ** initialization forms for bindings are not in scope of free special
1388 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1389 * incompatible change: thread support for non-NPTL systems has
1390 been removed - locking is buggy and unreliable. A threaded
1391 SBCL build will now warn at startup and refuse to create
1392 new threads, unless futex support is detected in the kernel
1393 * incompatible change: the top level REPL now has only an ABORT
1394 restart associated with it, not TOPLEVEL and ABORT as it used to.
1395 TOP and TOPLEVEL are now available as debugger commands for
1396 returning to the top level.
1397 * incompatible change: forms evaluated in the REPL now use the
1398 global optimization policy.
1399 * incompatible change: user- and system-initialization files are
1400 no longer processed with LOAD, but by READ and EVAL; hence the
1401 global optimization policy, startup package, readtable, etc,
1403 * The .fasl file format number has been incremented because of
1404 various incompatible changes.
1405 * internal entry point details and argument counts no longer appear
1406 in backtraces unless explicitly requested by setting
1407 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1408 * built-in and standard functions no longer have names like "top
1409 level local call to FOO".
1410 * fixed bug 32: functions defined in non-null lexical environments
1411 now have more legible printed representation
1412 * fixed bug 33: functions defined in non-null lexical environemnts
1413 are now more amenable to inspection by INSPECT.
1414 * workaround for bug 354: XEPs no longer appear in backtraces unless
1415 explicitly requested.
1416 * fixed bug: receiving the signal which results from attempting to
1417 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1418 SIGBUS on other *BSDs) on architectures where the C stack is also
1419 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1420 to debugger confusion or wild instability. (reported by Cheuksan
1422 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1423 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1424 (reported by Lutz Euler)
1425 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1426 compiled code. (reported by Paul Dietz)
1427 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1428 specialized parameter to a method to something that is not TYPEP
1429 the specializer is now possible.
1430 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1431 face of package deletion.
1432 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1433 pathname. (reported by David Tolpin on c.l.l)
1434 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1435 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1436 STRING-STREAM and FILE-STREAM, respectively.
1437 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1438 than previously; in addition, its test suite is now run on build.
1439 (thanks to Luke Gorrie)
1440 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1441 * optimization: added a immediate representation for single-floats
1443 * fixed some bugs revealed by Paul Dietz' test suite:
1444 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1445 ** attempting to create a package with a colliding nickname causes
1446 correctable errors to be signalled.
1447 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1448 ** several x86-64 backend bugs related to sign-extension of immediate
1451 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1452 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1453 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1455 * fixed loading of multiply forward-referenced layouts.
1456 (thanks to Cheuksan Wang)
1457 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1458 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1459 (see COMPILE-FILE documentation for details of :PRINT option)
1460 * fixed bugs 19 and 317: fixed-format floating point printing is
1461 more accurate. This also fixes a bug reported by Adam Warner
1462 related to the ~@F format directive.
1463 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1465 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1466 dispatch table if the from-char is a dispatch macro character.
1467 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1468 by the wordsize no longer ignore the last word. (reported by Lutz
1470 * fixed bug in type checking of dynamic-extent variables. (reported
1472 * optimization: sequence traversal functions use their freedom to
1473 coerce function designators to functions.
1474 * optimization: code with many calls to CLOS methods specialized on
1475 CLOS classes has had redundant type checks removed.
1476 * optimization: added declarations to speed up operations that access
1477 the internal character database (for example STRING-UPCASE)
1478 * optimization: comparison operations between floats and sufficiently small
1479 fixnums no longer create extra rationals
1480 * fixed some bugs related to Unicode integration:
1481 ** portions of multibyte characters at the end of buffers for
1482 character-based file input are correctly transferred to the
1483 start of the buffer at the next read.
1484 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1485 passing it through to OPEN.
1486 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1487 argument given to internal calls to OPEN.
1488 ** the built-in comment readers (introduced by character sequences
1489 ";" and "#|") are more forgiving to encoding errors; they will
1490 STYLE-WARN and then attempt to resync the stream at a character
1491 boundary. (thanks to Teemu Kalvas)
1492 * fixed some bugs in the x86-64 port:
1493 ** Negative short int return values from c-calls are sign-extended
1495 ** The stack is aligned to 16-bytes for c-calls, as required by
1496 the ABI. (thanks to Cheuksan Wang)
1497 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1498 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1499 for enums. (thanks to Vincent Arkesteijn)
1500 ** Multiple small optimizations and bugfixes for floating point
1502 * fixed some bugs revealed by Paul Dietz' test suite:
1503 ** Space, Tab, Linefeed, Return and Page have the invalid
1504 secondary constituent character trait.
1505 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1507 ** WITH-INPUT-FROM-STRING should only update the index place on
1509 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1510 works more reliably.
1511 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1512 with one-dimensional arrays with fill pointers.
1513 ** TYPE-OF failed on a complex with an integer realpart and a
1515 ** compiler failure during type inference for the code of form
1516 (IF (EQL X (THE ...)) ...) (MISC.535).
1518 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1519 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1520 platform. The port supports 61-bit fixnums, large memory spaces
1521 and reloading shared object files.
1522 * enhancement: saving cores with foreign code loaded is now
1523 supported on ppc/Darwin in addition to the previously supported
1525 * enhancement: the statistical profiler now walks deeper into the
1526 call stack for more meaningful call-graphs and accrued time
1527 reports (x86/x86-64 only). It also now reports time spent in
1529 * enhancement: it is now possible to trace most individual methods
1530 of a generic function in addition to tracing the generic function
1532 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1534 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1535 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1536 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1537 produces an error. (thanks to Vincent Arkesteijn)
1538 * bug fix: NAMESTRING on pathnames with :WILD components in their
1539 directories works correctly. (thanks to Artem V. Andreev)
1540 * fixed bug 125: compiler preserves identity of closures. (reported
1542 * bug fixed: functions with &REST arguments sometimes failed with
1543 "Undefined function" when compiled with (DEBUG 3). (reported by
1545 * bug fix: overflow during compiling of setting element of a bit
1546 vector with constant index and value. (reported by Timmy Douglas)
1547 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1548 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1549 * on x86 compiler partially supports stack allocation of dynamic-extent
1551 * GO and RETURN-FROM do not check the extent of their exit points
1552 when compiled with SAFETY 0.
1553 * fixed some bugs related to Unicode integration:
1554 ** encoding and decoding errors are now much more robustly
1555 handled; it should now be possible to recover even from invalid
1556 input or output to the terminal. (thanks to Teemu Kalvas)
1557 ** provided a first cut at implementing STRING-TO-OCTETS and
1558 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1559 ** altered the SB-MD5 contributed module slightly, changing the
1560 interface just enough for it to be supportable for builds where
1561 lisp characters are not eight bits.
1562 * fixed some bugs revealed by Paul Dietz' test suite:
1563 ** the FORMATTER-generated functions for ~V[ conditionals require
1564 the correct number of arguments.
1565 ** READ-FROM-STRING returns the mandated second value when applied
1566 to displaced strings.
1567 ** the #\Rubout and #\Backspace characters are treated as invalid
1568 constituent characters by the tokenizer.
1570 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1571 * new feature: reloading changed shared object files with
1572 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1573 * new feature: references to foreign variables and functions
1574 can now be compiled and loaded before the corresponding shared
1575 object file is loaded, as long as the foreign definitions are
1576 available at runtime.
1577 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1578 supported. (thanks to Dan Debertin)
1579 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1580 just keywords. (thanks to Vincent Arkesteijn)
1581 * fixed bug #331: structure-class instances corresponding to
1582 DEFSTRUCT forms are now created eagerly.
1583 * fixed bug #345: backtraces from calls to undefined functions work
1584 on x86 as well. Related bug #61 is now also partially fixed on x86
1585 (backtraces from throws to unknown catch tags.)
1586 * bug fix: lambda-list parsing is now stricter vrt. order and number
1587 of lambda-list keywords.
1588 * bug fix: as specified by AMOP, an error is signalled if a
1589 class-option appears multiple times in a DEFCLASS form. (reported
1591 * bug fix: RANDOM can be compiled when the compiler derives the type
1592 of its numeric argument as a disjoint set of small integers.
1593 (reported by Paul Dietz)
1594 * bug fix: starting a core saved with shared objects loaded when
1595 those objects are not available no longer causes threaded SBCL to
1596 hang. (reported by Sean Ross)
1597 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1599 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1600 argument, which must always be NIL. (reported by Kalle Niemitalo)
1601 * bug fix: printing 1.0d+23 no longer results in an error.
1602 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1603 * bug fix: structure slot setters preserve evaluation order. (thanks
1605 * fixed some bugs related to Unicode integration:
1606 ** RUN-PROGRAM can allow its child to take input from a Lisp
1607 stream. (reported by Stefan Scholl)
1608 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1609 locales. (reported by Ken Causey)
1610 * fixed some bugs revealed by Paul Dietz' test suite:
1611 ** INCF, DECF and REMF evaluate their place form as specified in
1613 ** FORMATTER expands ~{ ~} iteration directives with V or #
1614 parameters correctly.
1615 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1616 consequent uses no arguments correctly.
1617 ** the system has a partial understanding of the (COMPLEX RATIO)
1619 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1622 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1623 * new feature: a build-time option (controlled by the :SB-UNICODE
1624 keyword feature, enabled by default) for building the system with
1625 support for the entire 21-bit character space defined by the
1627 * new feature: the system now has rudimentary external-format
1628 support; the primary user-visible change at this time is that
1629 characters with the high bit set (such as lower-case-e-acute) will
1630 print correctly to a terminal in a UTF-8 environment.
1631 * minor incompatible change: BASE-CHAR no longer names a class;
1632 however, CHARACTER continues to do so, as required by ANSI.
1633 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1634 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1635 has been moved to the SB-EXT package (temporarily re-exported from
1637 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1638 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1640 * fixed bug #351: better error-handlind and reporting for malformed
1642 * fixed bug #350: bignum-printing is now more memory-efficient,
1643 allowing printing of very large bignums, eg. (expt 2 10000000).
1644 (reported by Bruno Haible)
1645 * fixed bug #302: better primitive-type selection for intersection
1647 * fixed bug #308: non-graphic characters now all have names, as
1648 required. (reported by Bruno Haible)
1649 * bug fix: redefining a class with different superclasses now correctly
1650 removes it from the direct-subclasses of its previous superclasses.
1651 (reported by David Morse)
1652 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1653 new value now works. (reported by Bruno Haible)
1654 * bug fix: correct canonicalization of multiple non-standard slot
1655 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1656 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1657 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1658 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1659 now exists, an signals an error.
1660 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1661 during macroexpansion for non-iteration variables as well. (reported
1662 by Bruno Haible for CMUCL)
1663 * bug fix: Cyclic structures and unprintable objects in compiler
1664 messages no longer cause errors. (reported by Bruno Haible)
1665 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1666 and READ-FROM-STRING all now return a primary value of NIL if
1667 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1668 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1669 (reported by Bruno Haible for CMUCL)
1670 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1671 do not designate a condition. (reported by Bruno Haible for
1673 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1674 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1675 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1676 permitting resolution in favour of any of the conflicting symbols.
1677 (reported by Bruno Haible for CMUCL)
1678 * bug fix: EQUAL compiler optimizations is less aggressive on
1679 strings which can potentially compare true despite having distinct
1680 specialized array element types.
1681 * bug fix: unit enumerations can be defined without dividing by
1682 zero. (thanks to Vincent Arkesteijn)
1683 * FORMAT compile-time argument count checking has been enhanced.
1684 (report from Bruno Haible for CMUCL)
1685 * a partial workaround for the bug 262: the compiler does not try to
1686 inline-expand a local function doing RETURN-FROM from a deleted
1687 BLOCK. (thanks to Peter Denno for the bug report and to David
1688 Wragg for the simple test case)
1689 * fixed some bugs revealed by Paul Dietz' test suite:
1690 ** RENAME-PACKAGE allows all package designators as new package
1692 ** constraint propagation and lambda variable substitution are
1693 more cautious in dealing with partially deleted code.
1694 ** compiler sometimes lost reoptimization passes.
1695 ** CERROR, given a condition as condition designator, passes its
1696 remaining arguments to the continue format control without
1698 ** Case-altering FORMAT directives work correctly on non-ASCII
1700 ** The REAL type specifier handles bounds outside the floating
1701 point ranges without signalling FLOATING-POINT-OVERFLOW.
1702 ** Functions with IR1-transformations can create intercomponent
1703 references to global functions.
1704 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1706 ** FORMAT ~R treats a nil value for its first parameter correctly.
1708 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1709 * enhancement: saving cores with foreign code loaded is now
1710 supported on x86/NetBSD and sparc/Linux in addition to the previously
1711 supported platforms.
1712 * bug fix: on some platforms repeated installations caused multiple
1713 copies of HTML documentation to be installed -- should not happen
1714 any more. (reported by Stefan Scholl)
1715 * bug fix: parsing self-recursive alien record types multiple times
1716 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1717 original patch by Helmut Eller for CMUCL)
1718 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1719 (thanks to Richard Kreuter)
1720 * bug fix: defining classes whose accessors are methods on existing
1721 generic functions in other (locked) packages no longer signals
1722 bogus package lock violations. (reported by François-René Rideau)
1723 * bug fix: special variables as DEFMETHOD parameters no longer have
1724 associated bogus type declarations. (reported by David Wragg and
1726 * bug fix: read-write consistency on streams of element-type
1727 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1728 * bug fix: redefiniton of the only method of a generic function with
1729 no DEFGENERIC no longer emits a full WARNING. In addition,
1730 redefinition of generic functions with no DEFGENERIC to an
1731 incompatible lambda list now signals an error. (thanks to Zach
1733 * bug fix: DEFGENERIC now works even when there's a function of the
1734 same name in an enclosing lexical environment. (thanks to Zach
1736 * fixed compiler failure, caused by instrumenting code during
1737 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1738 * optimization: added loop analysis and improved register allocation
1739 * fixed some bugs revealed by Paul Dietz' test suite:
1740 ** POSITION on displaced vectors with non-zero displacement
1741 returns the right answer.
1742 ** (SIMPLE-STRING) is a valid type specifier for sequence
1744 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1746 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1747 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1749 ** The justification version of the FORMAT ~< directive treats
1750 non-zero minpad parameter correctly.
1752 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1753 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1754 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1755 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1756 the supported interface.
1757 * new feature: Single-stepping of code compiled with DEBUG 2 or
1758 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1759 * new feature: saving cores with foreign code loaded is now
1760 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1761 Timothy Moore's work for CMUCL)
1762 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1763 parameters to * instead of NIL if no initform is supplied.
1764 (reported by Johan Bockgaard)
1765 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1766 correctly when FUNCALL appears as the car of the form. Note:
1767 despite this FUNCALL forms are not currently subject to
1768 compiler-macro expansion. (port of Raymond Toy's fix for the
1769 same from CMUCL, reported by Johan Bockgaard)
1770 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1771 (thanks for Teemu Kalvas)
1772 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1773 case instead of terminating the iteration (thanks for Julian
1774 Squires, Sean Champ and Raymond Toy)
1775 * bug fix: incorrect expansion of defgeneric that caused a style
1776 warning. (thanks for Zach Beane)
1777 * bug fix: slot accessor effective method computation works properly
1778 for classes with multiple non-standard applicable methods on
1779 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1780 * on x86 compiler supports stack allocation of results of LIST and
1781 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1782 CMUCL implementation by Gerd Moellmann)
1783 * fixed some bugs revealed by Paul Dietz' test suite:
1784 ** FORMAT strings with both the ~<~:;~> form of the justification
1785 directive and pretty-printing directives cause an error.
1787 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1788 * incompatible change: the internal functions
1789 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1790 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1791 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1792 instead of the old functions.
1793 * new feature: on platforms where "dladdr" is available foreign
1794 function names now appear in backtraces. (based on Helmut Eller's
1796 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1797 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1799 * documentation: ASDF manual is now distributed with SBCL.
1800 * bug fix: SBCL can now load its contributed modules with REQUIRE
1801 even if the system-provided entries have been removed from
1802 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1804 * bug fix: backtraces involving undefined functions or assembly
1805 routines are more informative. (thanks to Brian Downing)
1806 * bug fix: mutually referent alien structures now work correctly.
1807 (reported by Rick Taube)
1808 * bug fix: structures defined by WITH-ALIEN can be referred to
1809 within other definitions in the same WITH-ALIEN.
1810 * bug fix: division operators (MOD, TRUNCATE and the like) with
1811 constant zero divisors and integer dividends no longer generate
1813 * bug fix: provide default methods for INPUT-STREAM-P and
1814 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1815 * bug fix: improve the detection and resolution of MOP metacycles
1816 (where effective method computation depends on the generic function
1817 having its effective method computed). (reported by Bruno Haible)
1818 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1819 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1820 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1821 * bug fix: pathnames with relative directory components can be
1822 represented relative to default pathnames.
1823 * optimization: in taking the GCD of bignums, reduce the two bignums
1824 to approximately the same size (using Euclid's algorithm) before
1825 applying the more sophisticated binary GCD. (thanks to Juho
1827 * optimization: COUNT on bitvectors now operates word-at-a-time.
1828 * optimization: ASH with a positive, but not necessarily constant,
1829 (leftwards) shift, when in a modular context, is compiled to a
1831 * fixed some bugs revealed by Paul Dietz' test suite:
1832 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1833 if the corresponding argument is NIL.
1834 ** FORMAT directives accepting a minpad parameter treat negative
1836 ** literal commas as character parameters to FORMAT directives are
1838 ** literal spaces directly after ~<Newline> directives within a
1839 format-logical-block (~:< ~@:>) do not induce :FILL-style
1840 conditional newlines.
1841 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1842 as specified: it no longer includes conditional newlines).
1843 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1845 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1846 :PER-LINE-PREFIX argument does not evaluate to a string.
1848 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1849 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1850 the manual for details; for now, package locks can be disabled by
1851 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1852 if no major problems are found then it is likely that they will be
1853 compiled in unconditionally.
1854 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1855 unsupported operators on all platforms. To load a shared library
1856 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1857 object file, link it into a shared library outside of SBCL and
1858 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1860 * bug workaround: ROOM T and the GC were not working together
1861 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1862 reporting facility aren't preserved. That reporting has been
1863 disabled, so now until and unless someone figures out how to make it
1864 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1865 * minor incompatible change: as threatened around sbcl-0.8.0, the
1866 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1867 The symbols are also exported from SB-PCL for backwards
1868 compatibility, but more so than before SB-PCL should be treated as
1869 an implementation-internal package.
1870 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1872 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1873 It is known as of this release not to work on the Alpha, however.
1874 * fixed bug #167: errors signalled due to illegal syntax in method
1875 bodies are now more legible.
1876 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1877 designators and can hence be used with TYPEP.
1878 * fixed bug #333: CHECK-TYPE now ensures that the type error
1879 signalled, if any, has the right object to be accessed by
1880 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1881 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1882 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1884 * fixed bug #334: programmatic addition of slots using specialized
1885 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1886 and :ALLOCATION :CLASS slots.
1887 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1888 float exponents. (rereported by Peter Seibel)
1889 * fixed a bug: #\Space (and other whitespace characters) are no
1890 longer considered to be macro characters in standard syntax by
1891 GET-MACRO-CHARACTER.
1892 * fixed bug: initialization of condition class metaobjects no longer
1893 causes an instance of the condition to be created. (reported by
1895 * fixed bug: it is now possible to have slots such that
1896 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1897 description is neither :INSTANCE nor :CLASS.
1898 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1899 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1900 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1901 non-standard SLOT-DEFINITION-ALLOCATION values.
1902 * fixed bugs: various race conditions handling exiting threads.
1903 CL-PPCRE's thread torture test now passes.
1904 * fixed arguably-a-bug: GC time proportional to number of threads in
1905 system even when most of them are idle
1906 * optimization: improved performance of BIT and SBIT on bit-vectors.
1907 * .fasl file incompatibility: The fasl file version number has
1908 been incremented because of changes associated with package locks.
1910 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1911 * minor incompatible change: the system no longer provides
1912 optimization hints (conditions which are TYPEP
1913 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1914 mode; these hints are still emitted when higher SPEED optimization
1916 * new contrib module: a sampling profiler (profiling by statistical
1917 sampling, rather than by instrumenting functions) is available as
1918 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1919 * the behaviour of the standard function ED is now customizeable by
1920 third parties through a hook variable: see ED's documentation
1921 string for information on the protocol.
1922 * the compiler no longer emits efficiency notes for (FUNCALL X)
1923 when the type of X is uncertain under default optimization
1925 * fixed bug 276: mutating a binding of a specialized parameter to a
1926 method to something that is not TYPEP the specializer is now
1928 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1929 exhibit the required behaviour on the x86 platform. (thanks to
1930 Peter van Eynde, Eric Marsden and Bruno Haible)
1931 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1932 even for difficult arguments. (reported by Peter Graves)
1933 * fixed bug 141a: the backquote printer now descends quoted
1935 * fixed another bug in backquote printing: no more destructive
1936 modification of the form's list structure. (reported by Brian
1938 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1939 move between its address being taken and the call to
1940 interrupt_thread, fixing a crashing race condition.
1941 * the SB-POSIX contrib implementation has been adjusted so that it
1942 no longer exhibits ridiculously poor performance when constructing
1943 instances corresponding to C structs.
1945 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1946 * minor incompatible change: the sb-grovel contrib now treats C
1947 structures as alien (in the sense of SB-ALIEN) objects rather than
1948 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1949 has implications for memory management of client code
1950 (sb-grovel-returned objects must now be manually managed) and for
1951 type safety (alien objects now have full types).
1952 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1953 used to control emission of compiler diagnostics, rather than the
1954 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1955 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1956 quality should be considered deprecated.
1957 * install.sh now installs the user manual as well
1958 * (not quite a new documentable feature, but worth considering in
1959 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1960 beginnings of a semantically meaningful condition hierarchy is
1961 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1963 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1964 designator as the defaults argument.
1965 * fixed bug: Displaced arrays whose displaced-to array has become
1966 too small now cause ARRAY-DIMENSION to signal an error, providing
1967 for safer bounds-checking. (reported by Bruno Haible)
1968 * fixed bug: DEFCLASS slot definitions with identical :READER and
1969 :WRITER names now signal a reasonable error. (reported by Thomas
1971 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1973 * fixed bug: Invalid slot specification errors now print correctly.
1974 (thanks to Zach Beane)
1975 * fixed bug 320: Shared to local slot value transfers in class
1976 redefinitions now happen corrently from superclasses as
1977 well. (reported by Bruno Haible)
1978 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1980 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1981 improper lists correctly. (reported by Bruno Haible)
1982 * fixed bug 313: source-transform for <fun-name> was erroneously
1983 applied to a call of a value of a variable with name <fun-name>.
1984 (reported by Antonio Menezes Leitao)
1985 * fixed bug 307: The obsolete instance protocol ensures that
1986 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1987 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1988 remove dead unknown-values globs from the middle of the stack.
1989 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1990 easier to resume long computations after using *BREAK-ON-SIGNALS*
1991 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1992 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1993 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1994 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1995 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1996 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1998 * fixed a (fairly theoretical) bug in string printing: if
1999 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2000 does not have array-element-type equal to the most general string
2002 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2003 function name. (thanks to Nikodemus Siivola)
2004 * fixed bug: compile-time format string checker failed on
2005 non-closed ~{. (reported by Thomas F Burdick)
2006 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2007 don't warn on legal loop code involving more than one aggregate
2008 boolean. (thanks to Nikodemus Siivola)
2009 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2010 in keyword position to MAKE-PACKAGE was not being reported as
2012 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2013 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2015 * fixed bugs 280 and 312: the checking for multiple definitions in a
2016 file is less likely to become confused by uses of inline
2018 * fixed bug: the #S reader macro performs the keyword coercion
2019 specified for slot names. (reported by Kalle Niemitalo)
2020 * fixed bug: lambda lists may contain symbols whose names start with
2021 & but are not lambda-list-keywords; their occurrence triggers a
2023 * fixed bug 321: define-method-combination argument lambda lists do
2024 not cause invalid code to be generated when &optional or &aux
2025 variables have default values. (reported by Bruno Haible)
2026 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2027 have CLOS classes; this appears also to have fixed reported
2028 instability in stack exhaustion detection.
2029 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2030 MUFFLE-WARNING when no associated restart is present is now
2032 * optimization: rearranged the expansion of various defining macros
2033 so that each expands into only one top-level form in a
2034 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2036 * optimization: used a previously little-used slot in symbols to
2037 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2039 * fixed some bugs revealed by Paul Dietz' test suite:
2040 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2042 ** Fixed an optimization bug related to inheritance of initforms
2043 from local to shared slots.
2044 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2045 ** CAST optimizer forgot to flush argument derived type.
2046 ** print/read consistency on floats is now orders of magnitude
2047 more likely. (thanks also to Bruno Haible for a similar report
2049 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2050 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2051 for objects of type REAL. Make it so.
2052 ** SXHASH is specified (infelicitously) to respect similarity,
2053 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2054 it so. (thanks to Markus Ziegler)
2055 ** on the Alpha, the compiler succeeds in compiling functions
2056 returning a known number of arguments greater than 63.
2057 ** fixed handling of invalid NIL arguments in keyword position
2059 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2061 ** provided a readably-printable representation for RANDOM-STATE
2063 ** ensured that pathnames, where they have a namestring, always
2064 print using #P"..." syntax.
2066 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2067 * Support for the forthcoming 2.0 version of the NetBSD kernel
2068 running on x86 hardware has been added. (thanks to Perry
2069 E. Metzger most immediately, and others for their past work)
2070 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2071 current release version; SBCL's OpenBSD support had been broken
2072 since about the time of OpenBSD's switch to ELF binary format.)
2073 * [placeholder for DX summary]
2074 ** user code with &REST lists declared dynamic-extent, under high
2075 speed or space and low safety and debug optimization policy.
2076 * The manual has been converted to Texinfo format and the debugger
2077 chapter from the cmucl manual has been added.
2078 * A facility has been added to extract documentation strings from
2079 sbcl and store them as Texinfo-formatted snippets for inclusion in
2080 the manual (via Texinfo's @include directive)
2081 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2082 the test case to Dave Roberts)
2083 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2084 pointers. (reported by Sean Ross)
2085 * bug fix: PROFILE output is printed nicely even for large numerical
2086 values. (thanks to Zach Beane)
2087 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2088 greater than 32 handle EOF correctly.
2089 * bug fix: on X86 an immediate argument of the IMUL instruction is
2090 correctly printed in disassembly. (thanks to Lutz Euler)
2091 * bug fix: class slots in redefined classes preserve their old
2092 values. (thanks to Bruno Haible and Nikodemus Siivola)
2093 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2094 succeeds. (reported by Marco Baringer)
2095 * fixed some bugs revealed by Paul Dietz' test suite:
2096 ** READ-SEQUENCE now works on ECHO-STREAMs.
2097 ** RATIONALIZE works more according to its specification. (thanks
2100 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2101 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2102 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2103 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2104 should matter to you only if you rebind the printer control
2105 variables and then find you want different bindings in the
2106 debugger than in the ordinary execution of your program.)
2107 * The runtime build system has been tweaked to support building
2108 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2109 assembler and linker. (thanks to Nikodemus Siivola)
2110 * Unbound, undefined, undeclared variables now trigger full
2111 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2112 more often programmer error than deliberate exploitation of undefined
2114 * optimization: the hash algorithm for strings has changed to one
2115 that is less vulnerable to spurious collisions. (thanks to Juho
2117 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2118 needless bounds checking. (thanks to Juho Snellman)
2119 * optimization: implemented multiplication as a modular
2120 (UNSIGNED-BYTE 32) operation on the PPC backend.
2121 * fixed some bugs revealed by Paul Dietz' test suite:
2122 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2124 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2125 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2126 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2128 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2130 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2132 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2133 their output stream on EOF from read.
2134 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2135 have been read to end-of-file.
2136 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2138 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2139 description of determination of which consecutive characters
2141 ** Printing the "Space" character with escaping on now yields "#\\ ",
2142 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2143 ** Reading floating-point numbers with *READ-BASE* set to a number
2144 less than 10 works correctly.
2145 ** Reading floating-point numbers with *READ-BASE* set to a number
2146 more than 10 works correctly.
2147 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2148 the readtable currently in effect.
2150 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2151 * minor incompatible change: parsing of namestrings on a physical
2152 (Unix) host has changed; numbers after the final #\. in a
2153 namestring are no longer interpreted as a version field. This is
2154 intented to be largely invisible to the user, except that the
2155 meaning of the namestring "*.*.*" has changed: it now refers to a
2156 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2157 should usually be replaced by
2158 (make-pathname :name :wild :type :wild :version :wild)
2159 with the added benefit that this is more likely to be portable.
2160 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2161 option to OPEN now signals an error if the file being opened
2162 exists; this may have an impact on existing code.
2163 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2164 signals received, on the PowerPC platforms (both Linux and
2165 Darwin). (thanks to Pierre Mai for pointing out the location of
2167 * several fixes on OS X: The system now builds and runs cleanly on
2168 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2169 (thanks to Brian Mastenbrook)
2170 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2171 second-resolution: integer multiples of 1/3600 between -24 and 24.
2172 (thanks to Vincent Arkesteijn)
2173 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2174 type when called with 1 argument; PEEK-CHAR checked type of
2175 PEEK-TYPE only after having read first character from a
2176 stream. (reported by Peter Graves)
2177 * bug fix: the garbage collector now has much better locality
2178 behaviour, and in particular no longer treats all memory as being
2179 exhausted when many small objects point to each other in a deeply
2181 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2182 recognized as being TYPEP their class.
2183 * bug fix: the PUSHNEW documentation string has been corrected.
2184 (thanks to Vincent Arkesteijn)
2185 * bug fix: defaulting of the value for the last of an atypically
2186 large number of multiple values being bound was not being
2187 performed correctly on the Alpha or PPC platforms
2188 * optimization: implemented multiplication as a modular
2189 (UNSIGNED-BYTE 32) operation on the x86 backend.
2190 * optimization: SEARCH on simple-base-strings can now be open-coded.
2191 (see also contrib/compiler-extras.lisp for inspiration for
2192 teaching the compiler about the Boyer-Moore algorithm).
2193 * value, returned by MAX (and MIN) called with several EQUALP, but
2194 not EQL, arguments now does not depend on compiler settings.
2195 * fixed some bugs revealed by Paul Dietz' test suite:
2196 ** in stack analysis liveness information is propagated from
2197 non-local entry points.
2198 ** pathwise CAST removing failed when the CAST node did not start
2200 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2201 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2203 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2204 host is already defined.
2205 ** RENAME-FILE works on streams instead of signalling an internal
2207 ** PEEK-CHAR uses the current readtable when determining whether
2208 or not a character is whitespace.
2209 ** MERGE-PATHNAMES handles the case when the pathname does not
2210 specify a name while the default-pathname specifies a version.
2211 ** Pathnames now stand a better chance of respecting print/read
2213 ** Attempting to use standardized file system operators with a
2214 pathname with invalid :DIRECTORY components signals a
2216 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2218 ** DIRECTORY on logical pathnames is more correct.
2219 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2220 signal a TYPE-ERROR if their argument is not a stream.
2221 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2222 designator argument does not designate a stream.
2223 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2224 examining the synonym.
2225 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2227 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2228 element-type larger than ([UN]SIGNED-BYTE 32).
2230 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2231 * When built with the :SB-FUTEX feature, threaded builds now take
2232 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2233 for faster/more reliable mutex and condition variable support.
2234 * Incompatible change (but one you probably shouldn't have been using
2235 anyway): the interface and code for arbitrating between multiple
2236 threads in the same user session has been redesigned.
2237 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2238 error when called without an explicit environment argument.
2239 (thanks to Nikodemus Siivola)
2240 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2241 confused about their position. (thanks to Adam Warner and Gerd
2243 * bug fix: Pretty printing backquoted forms with unquotations in the
2244 argument list position of various code constructs such as LAMBDA
2245 now works correctly. (reported by Paul Dietz)
2246 * bug fix: Pretty printing unquotations no longer loses all
2247 stream position information.
2248 * optimization: performance of string output streams is now less
2249 poor for multiple small sequence writes.
2250 * optimization: performance of CSUBTYPEP in the presence of complex
2251 expressions involving CONS and NOT many times has been improved.
2252 (reported by Paul Dietz)
2253 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2255 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2256 streams for the REPL, for the debugger, and so on. (thanks to
2258 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2259 * fixed some bugs revealed by Paul Dietz' test suite:
2260 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2261 is ignored if it is too small, rather than propagating through
2262 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2263 ** extremely complex negations of CONS types were not being
2264 sufficiently canonicalized, leading to inconsistencies in
2266 ** VALUES tranformer lost derived type.
2268 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2269 * fixed a bootstrapping bug: the build process no longer assumes
2270 that the various BOOLE-related constants have the same value in
2271 host and target lisps. (noted by Paul Dietz' test suite on an
2272 SBCL binary built from CLISP)
2273 * The system can now be dynamically linked on the MIPS platform,
2274 which enables dynamic loading of foreign code from Lisp. (thanks
2275 to Ralf Baechle for discussions on the MIPS ABI)
2276 * The system now records debugging information for its own source
2277 files in a filesystem-position-independent manner, relative to
2278 the "SYS" logical host.
2279 * fixed a compiler bug: MV-LET convertion did not check references
2280 to the "max args" entry point. (reported by Brian Downing)
2281 * tweaked disassembly notes to be less confident about proclaiming
2282 some instruction as an LRA. (thanks to Brian Downing)
2283 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2284 now each have their own history, command character, and other
2285 characteristics. (thanks to David Lichteblau)
2286 * fixed some bugs revealed by Paul Dietz' test suite:
2287 ** compiler failure in compiling LOGAND expressions including a
2289 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2290 shift greater than 32.
2291 ** FLUSH-DEST did not mark blocks for type check regeneration.
2292 ** HANDLER-CASE failed to accept declarations in handler clauses
2293 in some circumstances.
2295 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2296 * New code in contrib/sb-introspect (still probably not entirely
2297 stable yet) provides some support for smart Lisp development
2298 environments like SLIME.
2299 * The conditions signalled for errors occurring when loading .fasl
2300 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2301 in a way which should help ASDF recover gracefully.
2302 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2303 stuff has been cleaned up. If you code contrib/ stuff, this might
2304 affect you, and you can look at contrib/README, contrib/STANDARDS,
2305 and/or the 0.8.4.27 diff to check.
2306 * In full calls the compiler now does not generate checks for declared
2307 argument types for all arguments.
2308 * various threading fixes
2309 ** and some experimental patches which didn't make it into
2310 the main tree for this release, but which are shipped in
2311 contrib/experimental-thread.patch as a possible fix for some
2312 failures (deadlock, spinning...) in GC-intensive multithreaded
2314 * fixed PPC build problem (source code incompatibility of different
2315 library versions): added offsetof() hackery which attempts to divine
2316 where glibc maintainers put uc_mcontext today
2317 * fixed bug 282: compiler does not trust type assertions while passing
2318 arguments to a full call.
2319 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2320 &OPTIONAL VALUES type parameter.
2321 * fix bug 214: algorithm for noting rejected templates is now more
2322 similar to that of template seletion. (also reported by rydis on
2324 * fixed bug 141b: printing backquoted information readably and prettily
2325 inserts a space where necessary.
2326 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2327 (CHARACTER 10) are now reported as errors, rather than propagated
2328 as unknown types. (reported by piso on #lisp)
2329 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2330 with values NIL and :ERROR. (thanks to Milan Zamazal)
2331 * fixed bug 191c: CLOS now does proper keyword argument checking as
2332 described in CLHS 7.6.5 and 7.6.5.1.
2333 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2334 longer raise an error; further, using a list as a for-as-arithmetic
2335 counter now raises a meaningful error.
2336 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2337 now understood by sequence creation functions such as MAKE-SEQUENCE
2339 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2340 a STRING-INPUT-STREAM.
2341 * compiler enhancement: SIGNUM is now better able to derive the type
2343 * type declarations inside WITH-SLOTS are checked. (reported by
2345 * fixed some bugs revealed by Paul Dietz' test suite:
2346 ** incorrect optimization of TRUNCATE for a positive first
2347 argument and negative second.
2348 ** compiler failure in let-convertion during flushing dead code.
2349 ** compiler failure while deriving type of TRUNCATE on an
2350 interval, containing 0.
2351 ** ASH of a negative bignum by a negative bignum count now returns
2353 ** intersection of CONS types now canonicalizes properly, fixing
2354 inconsistencies in SUBTYPEP.
2356 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2357 * incompatible change: The --disable-debugger command line
2358 option now clobbers the debugger at a more fundamental
2359 level, by redefining #'INVOKE-DEBUGGER instead of by
2360 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2361 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2362 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2363 enter the debugger REPL and then suffer recursive errors
2364 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2365 while under the new system, BREAK in --disable-debugger mode
2366 terminates the system just as an unhandled error would.
2367 * fixed compiler performance when processing loops with a step >1;
2368 * bug fix: DOCUMENTATION now retrieves generic function
2369 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2370 support has been systematized, and now supports the methods
2371 specified by ANSI, along with a default method and a method for
2372 slot documentation. (reported by Nathan Froyd)
2373 * bug fix: effective methods associated with a generic function are
2374 no longer cached over a change of that generic function's method
2375 combination. (reported by Andreas Fuchs)
2376 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2377 names a non-existent directory. (thanks to Andreas Fuchs)
2378 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2379 the PPC platform, lead to stack corruption; this has been fixed.
2380 (reported by Rainer Joswig)
2381 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2382 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2383 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2384 arithmetic in general, is now much more reliable on the PPC
2386 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2387 platform now returns the right answer.
2388 * optimization: restored some effective method precomputation in
2389 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2390 precomputation is now tunable.
2391 * optimization: compiler-internal data structure use has been
2392 reviewed, and changes have been made that should improve the
2393 performance of the compiler by about 20%.
2394 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2395 simple-base-strings and simple-bit-vectors is improved.
2396 * optimization: the optimization of 32-bit logical and arithmetic
2397 functions introduced in version 0.8.3 on the x86 has been
2398 implemented on the mips, ppc and sparc platforms; an
2399 implementation of the same facility, but for 64-bit arithmetic,
2400 has been added for the alpha.
2401 * microoptimization: the compiler is better able to make use of the
2402 x86 LEA instruction for multiplication by constants.
2403 * bug fix: in some situations compiler did not report usage of
2404 generic arithmetic in (SPEED 3) policy.
2405 * bug 145b fix: compiler used wrong type specifier while converting
2406 MEMBER-types to numeric.
2407 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2409 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2410 index argument now works properly on the Alpha platform.
2411 * bug fix: floating point exception treatment on the Alpha platform
2413 * bug fix: FILE-POSITION works much better on string input and
2414 output streams. (thanks to Nikodemus Siivola)
2415 * bug fix: many threading/garbage collection symptoms sorted.
2416 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2417 might be pseudo-atomic.
2418 * internal change: Stopping for GC is now done with signals not
2419 ptrace. GC is now done in whichever thread wanted it, instead of
2421 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2423 * bug fix: result form in DO is not contained in the implicit
2425 * incompatible change: ICR structure is changed; the value part of
2426 CONTINUATION is now called LVAR; corresponding functions are
2427 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2428 * added type deriver for ISQRT (thanks to Robert E. Brown).
2429 * bug fix: better support for loading from the command line when an
2430 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2432 * fixed some bugs revealed by Paul Dietz' test suite:
2433 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2434 ** ROUND and FROUND now give the right answer when given very
2435 small float arguments.
2436 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2438 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2439 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2440 ** LCM should return a non-negative integer.
2441 ** PARSE-INTEGER returned the index of a terminator instead of the
2442 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2443 ** PARSE-INTEGER returned an incorrect index being applied to a
2445 ** LCM with two arguments of 0 returns 0 rather than signalling
2447 ** unsigned addition of a 32-bit constant with the high bit set no
2448 longer causes an internal compiler error on the x86.
2449 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2450 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2451 with negative last argument.
2452 ** byte specifiers with zero size and position no longer cause
2453 an error during type derivation.
2454 ** bignum multiplication on the Alpha platform now returns the
2456 * porting: The system now builds on SuSE AMD64, although it still
2457 generates a 32-bit binary.
2458 * .fasl file incompatibility: The fasl file version number has
2459 been incremented (because of the changes to internal compiler
2460 data structures referred to above).
2462 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2463 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2464 more accurately, on the Darwin kernel running on PowerPC hardware.
2465 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2466 * Compiler code deletion notes now signal a condition of type
2467 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2468 an associated MUFFLE-WARNING restart.
2469 * The compiler now performs limited argument count validation of
2470 constant format strings in FORMAT, and where appropriate in ERROR,
2471 CERROR and WARN. (thanks to Gerd Moellmann)
2472 * New ASDF-INSTALL contrib can be used for automatic download and
2473 installation of third-party Lisp code from CCLAN or other sites
2475 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2476 which forces another thread to execute a function supplied by the
2478 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2479 now accept and act upon their :ELEMENT-TYPE keyword argument.
2480 (reported by Martin Atzmueller, Edi Weitz)
2481 * bug fix: FILE-POSITION now accepts position designators up to
2482 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2483 is the greater. (thanks to Patrik Nordebo)
2484 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2485 Kalle Olavi Niemitalo)
2486 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2487 printable. (reported by Eric Marsden)
2488 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2489 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2490 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2491 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2492 (thanks to Stig Sandoe)
2493 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2494 UNREAD-CHAR now correctly includes the unread character in the
2495 target sequence. (thanks to Gerd Moellmann)
2496 * bug fix in threaded builds: the system can now be suspended and
2497 resumed by shell job control with minimal disruption.
2498 * bug fixes in times and timezones >2038AD
2499 * better handling of "where is GNU make?" problem in build scripts
2500 (thanks to Nikodemus Siivola)
2501 * new optimization: inside a named function any reference to a
2502 function with the same name is considered to be a self-reference;
2503 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2504 optimization quality.
2505 * new optimization on x86: logical functions and + now have
2506 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2507 used when the result is truncated to 32 bits.
2508 * VALUES declaration is partially enabled.
2509 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2510 * fixed some bugs revealed by Paul Dietz' test suite:
2511 ** The system now obeys the constraint imposed by
2512 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2513 types form a lattice under type intersection.
2514 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2515 ** ASSOC now ignores NIL elements in an alist.
2516 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2517 and (1+ MOST-POSITIVE-FIXNUM) answers.
2518 ** The addition of a method with invalid qualifiers to a generic
2519 function does not cause an error to be signalled immediately;
2520 a warning is signalled, and the error is generated only on
2521 calling the generic function.
2522 * changed .fasl file version number, in part to add type codes for
2523 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2524 obscure ANSI requirements
2526 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2527 * fixed bug 148: failure to inline-expand a local function left
2528 garbage, confusing the compiler.
2529 * fixed bugs 3cd: structure slot readers perform type check if the
2530 slot can have an invalid value (i.e. it is either not initialized
2531 or can be written with a less specific slot writer).
2532 * bug fix: the compiler now traps array references to elements off
2533 the end of an array; previously, the bounds checking in some
2534 circumstances could go off-by-one.
2535 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2537 * type declarations for array element types now obey the description
2538 on the CLHS page "Declaration TYPE", as per discussions on
2539 sbcl-help around 2003-05-08. This means that a declaration
2540 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2541 declaration, all references to BAR will be asserted or assumed
2542 (with THE, so dependent on compiler policy) to involve objects of
2543 type FOO. Note that no such declaration is implied in
2544 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2545 * declared types of functions from the "Conditions"
2546 chapter. (reported by Paul Dietz)
2547 * bug fix: CERROR accepts a function as its first argument.
2548 * bug fix: NTH an NTHCDR accept a bignum as index
2549 arguments. (reported by Adam Warner)
2550 * optimization: character compare routines now optimize comparing
2551 against a constant character. (reported by Gilbert Baumann)
2552 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2553 argument now works properly on the MIPS platform.
2554 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2555 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2556 declared to have a functional type, does not cause an error
2558 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2559 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2560 * fixed bug: if last continuation of a deleted block has a
2561 destination, this destination should be deleted too. (reported by
2563 * fixed a bug in the bootstrap process: the host compiler's values
2564 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2565 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2566 test case from Patrik Nordebo)
2567 * improved the ability of the disassembler on the PPC platform to
2568 provide helpful disassembly notes.
2569 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2570 the class in more cases than previously.
2571 * bug fix: FILE-POSITION now understands :START and :END for
2572 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2573 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2574 8) data. (thanks to David Lichteblau)
2575 * bug fix: it is possible to add a method to a generic function
2576 without lambda list.
2577 * bug fix: reader failed to signal END-OF-FILE inside an
2578 object representation. (reported by Nikodemus Siivola)
2579 * fixed some bugs revealed by Paul Dietz' test suite:
2580 ** LAST and [N]BUTLAST should accept a bignum.
2581 ** condition slot accessors are methods.
2582 ** (VECTOR NIL) is a subtype of STRING.
2584 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2585 * minor incompatible change: some nonsensical specialized lambda
2586 lists (used in DEFMETHOD) which were previously ignored now signal
2588 * minor incompatible change: the system is now aware of the types of
2589 variables in the COMMON-LISP package, and will signal errors for
2590 most violations of these type constraints (where previously they
2591 were silently accepted).
2592 * minor incompatible change: COMPILE-FILE now uses the freedom
2593 afforded (ANSI 3.2.2.3) to use derived function types for
2594 functions defined in the same file. This also permits the system
2595 to warn on static type mismatches and function
2596 redefinition. (Currently it does not work with high DEBUG level.)
2597 * minor incompatible change: VALUES declaration is disabled.
2598 * When issuing notes, the compiler now signals a condition of type
2599 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2600 restart for use in user handlers. It is expected that the
2601 COMPILER-NOTE condition will eventually become a condition
2602 supertype to a hierarchy of note types, which will then be
2603 handleable in a similar fashion. However, at the moment, no such
2604 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2605 but it's an implementation detail, not a classification for the
2607 * Changes in type checking closed the following bugs:
2608 ** type checking of unused values (192b, 194d, 203);
2609 ** template selection based on unsafe type assertions (192c, 236);
2610 ** type checking in branches (194bc).
2611 * A short form of VALUES type specifier has ANSI meaning (it has
2612 increased the number of situations when SBCL cannot perform type
2614 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2615 keywords or constants is permissible.
2616 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2617 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2618 classes. (thanks to Antonio Martinez)
2619 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2620 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2621 argument to a pretty-print stream. Instead, it leaves any such
2622 operations to DESCRIBE-OBJECT methods.
2623 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2625 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2626 lambda lists. (thanks to David Lichteblau)
2627 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2628 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2629 (thanks to Gerd Moellmann)
2630 * fixed bug 246: increased compilation speed of long
2631 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2633 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2635 * DEFCONSTANT now throws a condition of type
2636 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2637 constant to a non-EQL value; CONTINUE and ABORT restarts
2638 respectively change and preserve the value.
2639 * fixed bug 63: The code walker, part of the implementation of CLOS,
2640 is now better at handling symbol macros.
2641 * bug fix: There is no longer an internal implementation type named
2642 CL:LENGTH. (reported by Raymond Toy)
2643 * bug fix: In macro-like defining macros/special operators the
2644 implicit block does not enclose the lambda list.
2645 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2646 enclosed in parentheses) are not suitable as type specifiers, and
2647 their use properly signals an error now.
2648 * bug fix: An argument count mismatch for a type specifier in code
2649 being compiled no longer causes an unhandled error at compile
2650 time, but signals a compile-time warning.
2651 * fixed simple vector readable printing
2652 * bug fix: DESCRIBE takes more care over whether the class
2653 precedence list slot of a class is bound before accessing it.
2654 (reported by Markus Krummenacker)
2655 * bug fix: FORMATTER can successfully compile pretty-printer format
2656 strings which use variants of the ~* directive inside.
2657 * bug fix: SEARCH now applies its TEST predicate to the elements of
2658 the arguments in the correct order. (thanks to Wolfhard Buss)
2659 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2660 in type inference. (thanks to Robert E. Brown)
2661 * bug fix: Reading in symbols with an explicit package name of ""
2662 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2663 * fixed some bugs revealed by Paul Dietz' test suite:
2664 ** NIL is now allowed as a structure slot name.
2665 ** Arbitrary numbers, not just REALs, are allowed in certain
2666 circumstances in LOOP for-as-arithmetic clauses.
2667 ** Multiple class redefinitions before slot access no longer
2668 causes a type error.
2669 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2670 association between the name and a class.
2671 ** Generic functions with non-standard method-combination and over
2672 five methods all of which return constants no longer return NIL
2673 after the first few invocations. (thanks to Gerd Moellmann)
2674 ** CALL-NEXT-METHOD with no arguments now passes the original
2675 values of the arguments, even in the presence of assignment.
2676 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2678 ** DISASSEMBLE works with closures and funcallable instances.
2679 ** ADD-METHOD now returns the generic function, not the new method.
2680 ** FIND-METHOD signals an error if the lengths of the specializers
2681 is incompatible with the generic function, even if the ERRORP
2683 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2684 which its argument is a member.
2685 ** DEFCLASS only redefines the class named by its class-name
2686 argument if that name is the proper name of the class;
2687 otherwise, it creates a new class.
2688 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2689 of the UNBOUND-SLOT condition to the name of the slot.
2690 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2691 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2692 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2693 SLOT-MAKUNBOUND in the specified fashion.
2695 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2696 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2697 cross-compilation host. As a consequence, we can now bootstrap our
2698 way up to SBCL starting with a bare gcc toolchain and human-readable
2699 source code (first the source to CLISP, then the source to SBCL).
2700 * A contributed module containing a partial implementation of the
2701 simple-streams interface has been included. (thanks to Rudi
2703 * A contributed module implementing the RFC1321 Message Digest
2704 Algorithm, known as MD5, has been included.
2705 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2706 no longer has any effect, as the code controlled by this feature
2707 has been deleted. (As far as we know, no-one has ever built using
2708 this feature, and its semantics were confused in any case).
2709 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2710 work on conditions (as required by the ANSI specification),
2711 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2712 expected behaviour on conditions. Users should note, however,
2713 that such behaviour is not required by the ANSI specification,
2714 and so use of this behaviour may render their code unportable.
2715 * fixed some bugs revealed by Paul Dietz' test suite:
2716 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2718 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2719 ** SXHASH on condition objects no longer returns NIL.
2720 ** :ALLOCATION :CLASS slots are better treated; their values are
2721 updated on class redefinition, and initforms inherited from
2722 superclasses are applied.
2723 ** REMOVE-METHOD returns its generic function argument even when
2724 no method was removed.
2725 ** SHARED-INITIALIZE now initializes the values of the requested
2726 slots, including those with :ALLOCATION :CLASS.
2727 ** ALLOCATE-INSTANCE now works on structure classes defined via
2728 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2730 ** SLOT-EXISTS-P now works on conditions, as well as structures
2732 ** MAKE-LOAD-FORM now has the required methods on
2733 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2734 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2735 keyword, and now implements the SLOT-NAMES argument.
2736 ** methods with &OPTIONAL arguments no longer allow too many
2737 arguments to be passed in the call without error.
2738 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2739 option is consistent with the required arguments of the generic
2740 function lambda list.
2741 * bug fix: REQUIRE accepts a string designator. (Thanks to
2743 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2744 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2745 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2746 * bug fix: The compiler now checks for duplicated variables in macro
2748 * bug fix: SETQ on globals returns the correct value.
2749 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2750 not a COMPILER-ERROR (followed by some other strange error on
2751 choosing the CONTINUE restart).
2752 * bug fix: make.sh and friends are now more consistent in the way that
2753 they look for GNU "make".
2755 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2756 * experimental native threads support (on x86 Linux >=2.4 only).
2757 This is not compiled in by default: you need to add :SB-THREAD to
2758 the target features. See the "Beyond ANSI" chapter of the manual
2760 * fix for longstanding nonANSIism: The old distinction between
2761 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2762 The return value from CL:FIND-CLASS is now a CLOS class, and
2763 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2764 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2765 * An interface to the MetaObject Protocol, as described in Kiczales,
2766 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2767 MIT Press, 1991, is available from the SB-MOP package.
2768 * incompatible change: the SB-PCL package should now be considered
2769 a private implementation detail, and no longer a semi-private MOP
2771 * minor incompatible change: due to rearrangement for threads, the
2772 control stack and binding stack are now allocated at arbitrary
2773 addresses instead of being hardcoded per-port. Users affected by
2774 this probably have to be doing advanced things with shared
2775 libraries, and will know who they are.
2776 * minor incompatible change: Previously, all --eval forms used were
2777 processed with READ before any of them were processed with EVAL.
2778 Now each --eval form is processed with both READ and EVAL before
2779 the next --eval form is processed. (Thus package operations like
2780 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2781 work as the user might reasonably expect.)
2782 * minor incompatible change: *STANDARD-INPUT* is now only an
2783 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2785 * minor incompatible change: Y-OR-N-P is now character-oriented,
2786 not line oriented. Also, YES-OR-NO-P now works without errors.
2787 (thanks to Antonio Martinez)
2788 * sb-aclrepl module improvements: an integrated inspector, added
2789 repl features, and a bug fix to :trace command.
2790 * Known functions, which cannot be open coded by the backend, are
2791 considered to be able to check types of their arguments. (fixing
2792 a bug report by Nathan J. Froyd)
2793 * fixed a bug in computing method discriminating functions: It is
2794 now possible to define methods specialized on classes which have
2795 forward-referenced superclasses. (thanks to Gerd Moellmann)
2796 * fixed evaluation order in optional entries (reported by Gilbert
2798 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2799 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2800 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2801 required-by-AMOP NAME argument, as well as CLASS and
2802 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2803 * fixed bug 20: DEFMETHOD can define methods using names that are
2804 not the proper names of classes to designate class specializers.
2805 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2806 Unix file descriptors, instead of blowing up. (thanks to Antonio
2808 * Garbage collection refactoring: user-visible change is that a
2809 call to the GC function during WITHOUT-GCING will not do garbage
2810 collection until the end of the WITHOUT-GCING. If you were doing
2811 this you were probably losing anyway.
2812 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2813 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2814 TYPEP the latter but not the former.
2815 * The compiler issues a full WARNING for calls to undefined functions
2816 with names from the CL package.
2817 * MAP-INTO for a vector destination is open coded. (reported by
2818 Brian Downing on c.l.l)
2819 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2820 documentation string.
2821 * fixed some bugs revealed by Paul Dietz' test suite:
2822 ** COPY-ALIST now signals an error if its argument is a dotted
2824 ** Condition slots are now accessed more correctly in the presence
2825 of multiple initargs for a given slot.
2826 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2827 exclude restarts of the same name associated with a different
2829 ** DEFCLASS of forward-referenced classes with another
2830 forward-referenced class in the superclasses list no longer
2832 ** Condition slots are now initialized once each, not multiple
2833 times. (thanks to Gerd Moellmann)
2834 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2836 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2837 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2838 arguments contain duplicated elements.
2839 ** RESTART-CASE understands local macros.
2840 ** RESTART-CASE associates exactly its own restarts with a condition.
2841 ** ENDP in safe mode checks its argument to be of type LIST.
2842 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2843 in question is unbound.
2844 ** Optimized MAKE-INSTANCE functions no longer cause internal
2845 assertion failures in the presence of duplicate initargs.
2846 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2847 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2848 propagated even in the ctor optimized implementation of
2850 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2852 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2853 * a better implementation of SXHASH on (simple) bit vectors,
2854 measured both in execution speed and in distribution of results
2855 over the positive fixnums, has been installed. Likewise, a better
2856 implementation of EQUAL for simple bit vectors is now available.
2857 * fixed CEILING optimization for a divisor of form 2^k.
2858 * fixed bug 240 (emitting extra style warnings "using the lexical
2859 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2860 by Antonio Martinez)
2861 * fixed SXHASH, giving different results for NIL depending on type
2862 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2863 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2864 variable. (found by Rolf Wester)
2865 * SBCL does not ignore type declarations for special
2866 variables. (reported by rif on c.l.l 2003-03-05)
2867 * some bug fixes in contrib/sb-aclrepl/
2868 * fixed some bugs revealed by Paul Dietz' test suite:
2869 ** a bug in the CONS type specifier, whereby the CAR and CDR
2870 types got intertwined, has been fixed;
2871 ** the type system is now able to reason about the interaction
2872 between INTEGER and RATIO types more completely;
2873 ** APPEND, [N]REVERSE and NRECONC check that those their
2874 arguments, which must be proper lists, are really so;
2875 ** An array specialized to be unable to hold elements has been
2876 implemented, as required -- yes, really -- by ANSI;
2877 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2878 on malformed property lists;
2880 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2881 * incompatible packaging change: in line with Unix convention,
2882 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2883 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2884 * REQUIRE and PROVIDE are now optionally capable of doing something
2885 useful. See the documentation string for REQUIRE.
2886 * infrastructure for a managed SBCL contrib system: contributed
2887 modules in this release include:
2888 ** the ASDF system definition facility;
2889 ** an interface to the BSD Sockets API;
2890 ** an ACL-like convenience interface to the repl;
2891 (thanks to Kevin Rosenberg)
2892 ** an implementation of ROTATE-BYTE, with efficient implementation
2894 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2895 gives a closer approximation to the right answer than previously.
2896 (thanks to Raymond Toy)
2897 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2898 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2899 optional environment arguments, as required by ANSI.
2900 * fixed bugs in other functions taking environment objects, allowing
2901 calls with an explicit NIL environment argument to be compiled
2903 * fixed bug 228: primary return values from
2904 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2905 COMPILE or FUNCTION.
2906 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2907 :TYPE will no longer signal errors on innocuous objects.
2908 * fixed bug 231b: SETQ is better at respecting type declarations in
2909 the lexical environment.
2910 * fixed a bug in DEFCLASS: classes named by symbols with no or
2911 unprintable packages can now be defined.
2912 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2913 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2914 * fixed compiler failure related to checking types of functions.
2915 (reported by Robert E. Brown)
2916 * the compiler is now much more consistent in its error-checking
2917 treatment of bounding index arguments to sequence functions: in
2918 (SAFETY 3) code, errors will be signalled in almost all cases if
2919 invalid sequence bounding indices are passed to functions defined
2920 by ANSI to operate on sequences.
2921 * fixed a bug in the build procedure: documentation of SBCL-specific
2922 packages is now preserved and available in the final Lisp image.
2923 * lifted FDEFINITION lookup out of loops in the implementation of
2924 many list operations. (thanks to Robert E. Brown)
2925 * fixed a bug in the reader: the #n# reader macro now works for
2926 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2927 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2928 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2929 E. Brown and Tony Martinez respectively)
2930 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2931 SLOT-BOUNDP: in particular, it is now able to optimize them much
2932 better, and is now not vulnerable to having packages renamed.
2933 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2934 always called when a slot is not present in an instance. (thanks
2936 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2937 longer signal an error on unreading a character following EOF on
2938 the previous constituent stream. (thanks to Tony Martinez)
2939 * fixed some bugs revealed by Paul Dietz' test suite:
2940 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2941 not just nonnegative fixnums;
2942 ** the logical bit-array operators such as BIT-AND now accept an
2943 explicit NIL for their "opt-arg" argument (to indicate a
2944 freshly-consed result bit-array);
2945 ** ELT now signals an error on an invalid sequence index in safe
2947 ** the type system is now cleverer about negations of numeric
2948 types, and consequently understands the BIGNUM and RATIO types
2950 ** the type system is now cleverer about the interaction between
2951 INTEGER and RATIO types: while bugs still remain, many more
2952 cases are accurately computed;
2953 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2954 if it is in the last clause;
2955 ** CONSTANTLY now correctly returns a side-effect-free function in
2957 ** DECLARE is no longer treated as a special-operator; in
2958 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2959 * incremented fasl file version number due to the change in the
2960 DEFSTRUCT-SLOT-DESCRIPTION structure.
2962 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2963 * minor incompatible change: code processed by the "interpreter" or
2964 EVAL now has a compilation optimization policy of (DEBUG 2)
2965 (changed from (DEBUG 1)) to improve debuggability of interactive
2966 development, and to allow the use of the debug RETURN command in
2968 * an experimental implementation of the RETURN command for the
2969 debugger has been included. (thanks to Frederik Kuivinen)
2970 * fixed bug 62: constraints were not propagated into a loop.
2971 * fixed bug in embedded calls of SORT (reported and investigated by
2973 * fixed some bugs revealed by Paul Dietz' test suite:
2974 ** printing and reading of arrays with some dimensions having
2975 length 0 (thanks to Gerd Moellmann);
2976 ** BOA constructor with &AUX argument without a default value does
2977 not cause a type error;
2978 ** CONSTANTP now returns true for all self-evaluating objects.
2980 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2981 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2982 accessors that are related by inheritance, as specified in the
2983 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2984 Valtteri Vuorikoski)
2985 * The compiler is now able to inline functions that were defined in
2986 a complex lexical environment (e.g. inside a MACROLET).
2987 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2989 * Support for the upcoming FreeBSD-5.0 release has been included.
2990 (thanks to Dag-Erling Smorgrav)
2991 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2992 effect when it is not in a toplevel context.
2993 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2994 stand a better chance of being correct. (thanks to Gerd
2996 * fixed bug in COERCE, which now signals an error on coercing a
2997 rational to a bounded real type which excludes the expected
2999 * The compiler is now able to derive types more accurately from the
3000 COERCE and COMPILE functions.
3001 * fixed bug 223: functional binding is considered to be constant
3002 only for symbols in the CL package.
3003 * fixed bug 231: SETQ did not check the type of a variable being set
3004 (reported by Robert E. Brown)
3005 * A new optimization for MAKE-INSTANCE has been included, fixing
3006 various bugs (including relating to :ALLOCATION :CLASS slots and
3007 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3009 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3010 ** As required by ANSI, LOOP now disallows anonymous collection
3011 clauses such as COLLECT I in conjunction with aggregate boolean
3012 clauses such as THEREIS (= I 1);
3013 ** LOOP now signals an error when any variable is reused in the
3014 same loop (including the potentially useful construct analogous
3015 to WITH A = 1 WITH A = (1+ A);
3016 ** IT is only a special loop symbol within the first clause of a
3017 conditional loop clause;
3018 ** LOOP with a typed iteration variable over a hashtable now
3019 signals a type error iff it should.
3020 * fixed some other bugs revealed by Paul Dietz' test suite:
3021 ** FILE-STREAM now names the class previously known as FD-STREAM;
3022 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3023 argument) no longer signals an error;
3024 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3025 of the slot symbol, rather than using the current package
3026 ((:CONC-NAME "") continues to intern the slot's name in the
3028 * incremented fasl file version number, because of the incompatible
3029 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3030 of the new implementation of DEFINE-COMPILER-MACRO.
3032 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3033 * Support for building SBCL for MIPS platforms running in
3034 little-endian mode has now been checked in, and basic
3035 functionality on said platforms verified.
3036 * minor incompatible change: PCL now records the pathname of a file
3037 in which methods and the like are defined, rather than its
3039 * minor incompatible change: TRUENAME now considers the truename of
3040 a file naming a directory to be the pathname with :DIRECTORY
3041 component indicating that directory.
3042 * minor incompatible change: a NAMED clause in the extended form of
3043 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3044 reason for the previous behaviour is unclear.
3045 * more systematization and improvement of CLOS and MOP conformance
3046 in PCL (thanks to Gerd Moellman and Pierre Mai):
3047 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3049 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3050 primary methods with no specializers;
3051 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3053 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3054 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3055 FORWARD-REFERENCED-CLASSes; error reporting on
3056 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3058 ** SXHASH on CLOS instances now uses a slot internal to the
3059 instance to return different numbers on distinct instances,
3060 while preserving the same return value through invocations of
3062 ** DEFMETHOD signals errors when methods with longer incongruent
3063 lambda lists are added to generic functions;
3064 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3065 CLASS, as specified in AMOP;
3066 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3067 on the order returned by the primary method for classes of
3068 class STANDARD-CLASS;
3069 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3070 * fixed some bugs shown by Paul Dietz' test suite:
3071 ** DOLIST puts its body in TAGBODY;
3072 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3074 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3075 value producing form;
3076 ** if more variables are given to PROGV than values, extra
3077 variables are bound and made to have no value;
3078 ** NSUBSTITUTE on list arguments gets the right answer with
3080 ** ELT signals an error of type TYPE-ERROR when the index argument
3081 is not a valid sequence index;
3082 ** LOOP signals (at macroexpansion time) an error of type
3083 PROGRAM-ERROR when duplicate variable names are found;
3084 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3085 ** FUNCALL of special-operators now cause an error of type
3087 ** PSETQ now works as required in the presence of side-effecting
3088 symbol-macro places;
3089 ** NCONC accepts any object as its last argument;
3090 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3092 ** loop-for-as-package does not require a package to be explicitely
3094 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3096 * fixed bug 166: compiler preserves "there is a way to go"
3097 invariant when deleting code.
3098 * fixed bug 172: macro lambda lists with required arguments after
3099 &REST arguments now cause an error to be signalled. (thanks to
3101 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3102 bug. (thanks to Matthew Danish)
3103 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3105 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3106 arguments are lexically rebound. (thanks to Gerd Moellmann and
3108 * fixed bug 194: error messages are now more informative when there
3109 is no primary method applicable in a call to a generic
3110 function. (thanks to Gerd Moellmann)
3111 * fixed bug in command line argument checking (thanks to Julian
3113 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3114 (thanks to Matthew Danish)
3115 * incremented fasl file version number, because of the
3116 SXHASH-related changes in the layout of CLOS data structures
3118 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3119 * minor incompatible change: The runtime (the Unix executable named
3120 "sbcl") is now much pickier about the .core files it will load.
3121 Essentially it now requires .core files to descend from the same
3122 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3123 as the runtime does. (The intent is to prevent the crashes which
3124 can occur, and which can even be reported as mysterious failures,
3125 when people patch the sources or change the build parameters
3126 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3127 sbcl and .core files.)
3128 * fixed bug: VALUES-LIST is no longer optimized away.
3129 * fixed bug 142: The FFI conversion of C string values to Lisp
3130 string values no longer conses excessively. (thanks to Nathan
3131 Froyd porting Raymond Toy's fix to CMU CL)
3132 * began to systematize and improve MOP conformance in PCL (thanks to
3133 Nathan Froyd, Gerd Moellman and Pierre Mai):
3134 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3136 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3137 ** FINALIZE-INHERITANCE is now called on class finalization;
3138 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3139 argument precedence order.
3140 * fixed bug 202: The compiler no longer fails on functions whose
3141 derived types contradict their declared type.
3142 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3143 so it can be non-toplevel.
3144 * The fasl file version number has changed (because of the new
3145 implementation of DEFMACRO).
3146 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3147 safe code, that any length requirement by their type-specifier
3148 argument is valid. The exceptions to this are described in bug
3150 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3151 functions now check the types of their inputs as required by ANSI.
3152 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3153 introduced symbol is DECLAREd to be SPECIAL.
3154 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3155 * fixed bug: :COUNT argument to sequence functions may be negative
3156 * fixed bug: body of DO-SYMBOLS may contain declarations
3157 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3158 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3159 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3160 symbol macro only once
3161 * fixed printing of call frame when argument list is unavailable
3162 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3163 * compiler no longer signals WARNING on unknown keyword
3166 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3167 * A beta-quality port to the mips architecture running Linux,
3168 based on the old CMUCL backend, has been made. It has been tested
3169 on a big-endian kernel, and works sufficiently well to be able to
3170 rebuild itself; it has not been tested in little-endian mode.
3171 * fixed an inconsistency between gencgc.c and purify.c which made
3172 dumping/loading .core files unreliable
3173 * fixed bug 120a: The compiler now deals correctly with IFs where
3174 the consequent is the same as the alternative, instead of
3175 misderiving the return type. (thanks to Alexey Dejneka)
3176 * fixed bug 113: Logical pathnames are now dumpable (the logical
3177 host is resolved at load-time, throwing an error if it is not
3179 * fixed bug 174: FORMAT's error message is slightly clearer when a
3180 non-printing character is used in a format directive.
3181 * fixed several bugs in compiler checking of type declarations, i.e.
3182 violations of the Python "declarations are assertions" principle
3183 (thanks to Alexey Dejneka)
3184 * fixed several bugs in PCL's error checking (thanks to Gerd
3186 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3188 * fixed bug in compilation of functions as first class values
3189 (thanks to Antonio Martinez-Shotton)
3190 * The compiler's handling TYPE-ERRORs which it can prove will
3191 inevitably happen at runtime has been cleaned up and corrected
3192 in several ways. (thanks to Alexey Dejneka)
3193 * improved argument type checking for various basic arithmetic
3194 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3195 lavished on them in the past that they can be compiled in many
3196 ways in different special cases
3197 * fixed bug 181: compiler checks validity of user supplied type
3199 * cleaned up code flushing in optimization: Function calls which
3200 should signal errors for safety purposes (e.g. which ANSI says
3201 should signal errors when their arguments are of incorrect type)
3202 are no longer optimized away.
3203 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3204 * incremented fasl file version number, because changes in the
3205 implementation of sequence functions like COERCE caused
3206 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3207 in old inline expansions) to become undefined. (Actually these
3208 changes were later undone, so we might very well be binary
3209 compatible with 0.7.7 after all, but leaving the version number
3210 incremented seemed like the simplest and most conservative
3213 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3214 * An alpha-quality port to the parisc architecture running Linux,
3215 based on the old CMUCL backend, has been made. This, even more so
3216 than the other backends, should be considered still a work in
3217 progress; known problems include that the Linux kernel in 64-bit
3218 mode does not propagate the correct sigcontext structure to
3219 userspace, and consequently SBCL on a parisc64 kernel will not
3221 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3222 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3223 suppressed.) Also now that the compiler is looking at declarations
3224 in the environment, it checks optimization declarations as well,
3225 and suppresses inlining when (> DEBUG SPEED).
3226 * More fixes have been made to treatment of floating point exception
3227 treatment and other Unix signals. In particular, floating point
3228 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3229 * The detection and handling of control stack exhaustion (infinite
3230 or very deeply nested recursion) has changed. Stack exhaustion
3231 detection is now done by write-protecting pages at the OS level
3232 and applies at all optimization settings; when found, a
3233 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3234 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3235 be caught using IGNORE-ERRORS.
3236 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3237 correctly when transforming recursive LABELS functions to LETs.
3238 (thanks to Alexey Dejneka)
3239 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3240 that are names of constants or global variables.
3241 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3242 alien routines with docstrings.
3243 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3244 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3246 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3247 (thanks to Eric Marsden)
3248 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3249 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3250 * Bug fix: The SPARC backend can now compile functions involving
3251 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3252 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3253 to a functions expecting a PATHNAME-DESIGNATOR.
3254 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3255 lambda lists. (thanks to Alexey Dejneka)
3256 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3257 functions so that EXPORTing the name of the class doesn't cause
3258 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3259 to Antonio Martinez for reporting this)
3260 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3261 constant-folded without failing an assertion. (thanks to Einar
3262 Floystad Dorum for reporting this)
3263 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3264 types are now dealt with more correctly by the compiler.
3265 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3266 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3267 behaviour analogous to RENAME-FILE. This puts its behaviour more
3268 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3270 * The fasl file version number has changed again. (because of the
3271 bug fix involving the names of PCL MAKE-INSTANCE functions)
3273 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3274 * bug fix: Floating point exceptions are treated much more
3275 consistently on the x86/Linux and PPC/Linux platforms.
3276 * Array initialization with :INITIAL-ELEMENT is now much faster for
3277 cases when the compiler cannot open code the array creation, but
3278 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3279 array accesses have also seen a speed increase.
3280 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3281 specified. (This was at the root of some bad interactions between
3282 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3284 * bug fix: Internal error arguments for undefined functions are now
3285 computed correctly on the PPC/Linux platform.
3286 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3287 Raymond Toy's patch for CMU CL)
3288 * Support for the Solaris 9 operating environment has been included
3289 (thanks to Daniel Merritt)
3290 * A very ugly but hopefully complete draft of the missing FFI chapter
3291 of the manual has been created by reformatting the corresponding
3292 CMU CL manual chapter into (currently very ugly and incoherent)
3293 DocBook and bringing it up to date for SBCL behavior. Thus, the
3294 manual is now essentially complete, at least by my extreme
3295 once-and-only-once standards, whereby it's acceptable to refer to
3296 the doc strings of SB-EXT functions as the primary documentation.
3297 * The fasl file version number has changed again, due to cleanup of
3298 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3299 is no longer a static symbol.)
3301 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3302 * SBCL now builds with OpenMCL (version 0.12) as the
3303 cross-compilation host; also, more progress has been made toward
3304 bootstrapping under CLISP.
3305 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3307 * bug 158 fixed: The compiler can now deal with integer loop
3308 increments different from 1; fixing this turned out also to fix
3310 * bug 169 fixed: no more bogus warnings about using lexical bindings
3311 despite the presence of perfectly good SPECIAL declarations (thanks
3312 to David Lichteblau)
3313 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3314 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3315 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3317 * bug 180 fixed: Method combination specifications no longer ignore
3318 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3319 * bug fix: Structure type predicate functions now check their argument
3320 count as they should.
3321 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3322 correctly. (thanks to Pierre Mai)
3323 * minor incompatible change: The --noprogrammer option is deprecated
3324 in favor of the new --disable-debugger option, which is very similar.
3325 (The major difference is that it takes effect at a slightly different
3326 time at startup, causing handling of errors in --sysinit and
3327 --userinit files will be affected differently.) The
3328 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3329 been added to allow this functionality to be controlled from ordinary
3330 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3331 maintainers, who might want to run non-interactive scripts to
3332 build SBCL cores which will later be used interactively.)
3333 * minor incompatible change: The LOAD function no longer, when given
3334 a wild pathname to load, loads all files matching that pathname.
3335 Instead, an error of type FILE-ERROR is signalled.
3337 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3338 * bug 147 fixed: The compiler preserves its block link/count
3339 invariants more correctly now so that it doesn't crash. (thanks
3341 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3343 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3344 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3345 no-value-for-this-character as NIL (as specified by ANSI).
3346 * HOST-NAMESTRING on physical pathnames now returns a string that is
3347 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3348 (thanks to Christophe Rhodes)
3349 * The Alpha port handles icache flushing more correctly. (thanks to
3351 * More progress has been made toward bootstrapping under CLISP. (thanks
3352 to Christophe Rhodes)
3353 * The fasl file format has changed again, because dynamic loading
3354 on OpenBSD (which has non-ELF object files) motivated some cleanups
3355 in the way that foreign symbols are transformed and passed around.
3356 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3357 is no longer treated as whitespace by the reader, but instead as
3358 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3359 |A
\7fB|, instead of A as it used to.
3361 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3362 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3363 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3364 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3365 this as of 0.7.1.45, but was left out of the previous news section
3366 (thanks to Dan Barlow)
3367 * SBCL now runs on the Solaris operating system on SPARC architectures
3368 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3369 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3370 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3371 * SPARC backend cleanups, allowing builds of cores optimized for V8
3372 and V9 SPARCS, and also emission of code targeted to a particular
3373 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3375 * SBCL is closer to bootstrapping under CLISP, thanks to various
3376 fixes by Christophe Rhodes.
3377 * The fasl file format has changed again, to allow the compiler's
3378 INFO database to support symbol macros.
3379 * The user manual (in doc/) is formatted into HTML more nicely.
3380 (thanks to coreythomas)
3381 * The system is smarter about SUBTYPEP relationships, especially
3382 those involving NOT types (including types such as ATOM which are
3383 represented internally using NOT types). Thus SUBTYPEP is less
3384 likely to return (VALUES NIL NIL) in general, and in particular
3385 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3387 * The fasl file format has changed again, because the internal
3388 representation of types now includes a new slot to support the new
3389 SUBTYPEP-of-NOT-types logic.
3390 * (not a change in the main branch of SBCL, but a related prototype
3391 which can hopefully be merged into the main branch of SBCL in the
3392 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3393 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3394 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3395 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3396 dispatch macro character combinations. (thanks to Alexey Dejneka)
3397 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3398 that superficially look like logical namestrings correctly.
3399 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3401 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3402 default supplied-p) for &optional and &key arguments. (thanks to
3405 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3406 * incompatible change: The compiler is now less aggressive about
3407 tail call optimization, doing it only when (> SPACE DEBUG) or
3408 (> SPEED DEBUG). (This is an incompatible change because there are
3409 programs which relied on the old CMU-CL-style behavior to optimize
3410 away their unbounded recursion which will now die of stack overflow.)
3411 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3412 for non-GENCGC systems has been increased to 20M (since that
3413 seems much closer to the likely performance optimum for modern
3414 systems than the old 4M value was)
3415 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3416 and *DEBUG-PRINT-LEVEL*
3417 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3418 of CMU CL's support for SPARC, and various endianness and other
3419 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3420 * new syntactic sugar for the Unix command line: --load foo.bar is now
3421 an alternate notation for --eval '(load "foo.bar")'.
3423 ** The system now detects stack overflow and handles it gracefully,
3424 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3425 optimization settings. (This is a good thing in general, and
3426 its introduction in this version should be particularly timely
3427 for anyone whose code fails because of suppression of tail
3429 ** The system now hunts for the C variable "environ" in a more
3430 devious way, to avoid segfaults when the C library version differs
3431 between compile time and run time. (thanks to Christophe Rhodes)
3432 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3433 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3434 ** The compiler no longer issues bogus style warnings for undefined
3435 classes in the same source file as the DEFCLASSes which defined
3436 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3438 ** fixes in CONDITION class precedence list for undefined function
3439 errors (thanks to Alexei Dejneka)
3440 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3441 correctly. (thanks to Dan Barlow)
3442 ** portability fixes aiming at bootstrapping under CLISP (thanks
3443 to Dave McDonald and Christophe Rhodes)
3444 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3445 ** fixes in type translation and and type inference (thanks to
3447 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3448 ** various fixes in the new ports (thanks to Dan Barlow)
3449 * several changes related to debugging:
3450 ** suppression of tail recursion, as noted above
3451 ** stack overflow detection, as noted above
3452 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3453 is now the default. (For some time encapsulation has been more
3454 reliable than the breakpoint-based :ENCAPSULATE NIL
3455 implementation, at least on X86 systems; and I just noticed that
3456 encapsulation also seems closer to the spirit of the ANSI
3459 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3461 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3462 up properly again. (There was a packaging bug in 0.7.0 which
3463 left their definitions in SB-SYS::LOAD-FOREIGN and
3464 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3465 vital for most things which interface to C-level interfaces,
3466 like extensions working with sockets or databases or
3467 Perl-compatible regexes or whatever, and the need to fix
3468 this bug is the main reason that 0.7.1 was released so
3470 ** DEFGENERIC is now choosier about the methods it redefines, so that
3471 reLOADing a previously-LOADed file containing DEFGENERICs does
3472 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3473 cycle now works as it should. (thanks to Alexey Dejneka)
3474 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3475 fixed by Christophe Rhodes. (He actually submitted this patch
3476 months ago, and I delayed until after 0.7.0.)
3477 ** Bug 111 (internal compiler confusion about runtime checks on
3478 FUNCTION types) was fixed by Alexey Dejneka.
3479 * Some internal cleanups (getting rid of variables which aren't
3480 needed now that the byte interpreter is gone) caused the fasl
3481 file format number to change again.
3483 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3484 * major incompatible change: The default fasl file extension, i.e. the
3485 default extension for files produced by COMPILE-FILE, has changed
3486 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3488 ** There are many changes in the implementation of the compiler.
3489 SBCL is now essentially a compiler-only implementation of ANSI
3490 Common Lisp. EVAL still "interprets" a few special cases, but
3491 almost all the interesting cases are handled by creating
3492 a LAMBDA expression, calling COMPILE on it, then calling
3493 FUNCALL on the result.
3494 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3495 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3496 Since the code is newer, there might still be some new bugs
3497 (though not as many as before Martin Atzmueller's fixes:-). But
3498 the new code is substantially simpler and clearer, and hopefully
3499 any remaining bugs will be simpler, less fundamental, and more
3500 fixable then the bugs in the old code.
3501 ** The revised compiler is still a little unsteady on its feet.
3503 *** The debugging information it produces (particularly the names
3504 of FUNCTION objects) is sometimes much less useful than what
3505 the old compiler produced.
3506 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3507 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3508 within a MACROLET) has been temporarily weakened.
3509 ** There are new compiler optimizations for various functions:
3510 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3511 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3512 *** the math functions TRUNCATE, FLOOR, and CEILING
3513 *** the function-of-all-trades COERCE
3514 Mostly these should be transparent, but there's one
3515 potentially-annoying problem (bug 117): when the compiler
3516 inline-expands a function and does type analysis on the result,
3517 it can create control paths which have type mismatches, and
3518 when it can't prove that those control paths aren't taken,
3519 it will issue WARNINGs about the type mismatches. This is
3520 a particular problem in practice for the new sequence functions.
3521 It's not clear how this should be fixed, and for now, a
3522 workaround is given in the entry for 117 in the BUGS file.
3523 ** (Because of the interaction between the two previous items --
3524 occasional inlining problems and new inline expansions -- some
3525 of the new sequence function optimizations won't really kick in
3526 completely until debugging information, and then inlining, are
3527 straightened out in some future version.)
3528 * minor incompatible changes:
3529 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3530 DIRECTORY no longer implicitly promotes NIL slots of its
3531 pathname argument to :WILD. In particular, when you ask for the
3532 contents of a directory (which you used to be able to do without
3533 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3534 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3535 ** changes in behavior that ANSI explicitly defines to be
3536 implementation dependent:
3537 *** The new compiler-only implementation still conforms with ANSI,
3538 but acts a little different than before. Besides the obvious
3539 changes in performance tradeoffs (that the cost per form passed
3540 to EVAL has gone up, and the cost per form executed by EVAL
3541 has gone down), the behavior of the system changes a little
3542 because there are no longer any interpreted function objects.
3543 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3544 e.g. doing COMPILE on the output of interactive DEFUN is
3546 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3548 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3549 has changed from (:CL) to NIL.
3550 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3551 e.g. #\Newline and #\Tab, don't have names specified in the
3552 ANSI Common Lisp standard, is now based on their ASCII symbolic
3553 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3554 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3555 are no longer used for output.
3556 ** changes in internal implementation constants:
3557 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3558 4 million. (If your application spends a lot of time GCing and
3559 you have a lot of RAM, you might want to experiment with
3560 increasing it even more.)
3561 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3562 However, almost all old code should still continue to work without
3563 immediate update, as SB-C-CALL is now a (deprecated) nickname
3565 ** Old operator names in the style DEF-FOO are now deprecated in
3566 favor of new corresponding names DEFINE-FOO, for consistency with
3567 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3568 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3569 internal symbols, but a few supported extensions like
3570 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3571 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3572 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3573 etc. as you get deeper into recursive calls to the debugger
3574 command loop, instead of the old "5]", "5]]", "5]]]"
3575 sequence. (I was motivated to do this when squabbles between
3576 ILISP and SBCL left me very deeply nested in the debugger. In the
3577 short term, this change will probably provoke more ILISP/SBCL
3578 squabbles, but hopefully it will be an improvement in the long run.)
3579 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3580 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3581 function, since declaiming return types involving aliens is
3582 (1) annoyingly messy to do by hand and (2) vital to efficient
3583 compilation of code which calls such functions.
3584 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3585 longer reexported by the SB-EXT package. They're solely useful
3586 for alien code, so it seems more logical that you should get
3587 them from the SB-ALIEN package, not in SB-EXT.
3588 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3589 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3590 features. Instead, the code that they used to control is always
3591 built into the system.
3592 * many other bug fixes
3593 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3594 advantage of the new EVAL-WHEN stuff and to clean them up in
3595 general, and they are now more ANSI-compliant in a number of
3596 ways. Martin Atzmueller is responsible for a lot of this.
3597 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3599 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3600 *** correct ERROR type for various file operations
3601 *** some fixes for Lisp streams
3602 *** DEFMETHOD syntax checking
3603 *** changing old weird representation of debug information as
3604 strings (which, among their other deficiencies, don't transform
3605 correctly when you rename packages, and don't change their
3606 print representation when you change things like *PACKAGE*
3607 and *PRINT-LENGTH*) to symbols and lists of symbols
3608 He also made several improvements and fixed several bugs in DESCRIBE.
3609 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3611 *** misbehavior of WRITE-STRING/WRITE-LINE
3612 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3613 and several other LOOP problems as well
3614 *** DIRECTORY when similar filenames are present
3615 *** DEFGENERIC with :METHOD options
3616 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3617 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3618 *** argument ordering in FIND with :TEST option
3619 *** mishandled package designator argument in APROPOS-LIST
3620 *** various problems in the backquote readmacro
3621 *** a bug in APROPOS
3622 *** probably some others that I'm not describing very well here,
3623 since the CVS log documents them by reference to sbcl-devel
3624 messages, and the SourceForge archives aren't working well.:-(
3625 ** Dan Barlow improved the Alpha port (and is making progress on the
3626 PPC port, for those of you who think different).
3627 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3628 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3629 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3630 continues to work on the SPARC port (for those of you in a position
3631 to look down upon our little PC-compatible boxes from a great height).
3632 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3633 and uses that copy, rather than the current dynamic value, when
3634 it's trying to decide whether to truncate output. Thus e.g.
3635 (let ((*print-lines* 50))
3636 (pprint-logical-block (stream nil)
3638 (let ((*print-lines* 8))
3639 (print (aref possiblybigthings i) stream)))))
3640 should now truncate the logical block only at 50 lines, instead of
3641 often truncating it at 8 lines, as it did before.
3642 * The doc/cmucl/ directory, containing old CMU CL documentation
3643 from the time of the fork, is no longer part of the base system.
3644 SourceForge has shut down its anonymous FTP service, and with it
3645 my original plan for distributing the old CMU CL documentation
3646 there. For now, if you need these files you can download an old
3647 SBCL source release and extract them from it.
3648 * The fasl file version number changed again, for dozens of reasons,
3649 some of which are apparent above.
3651 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3652 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3653 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3654 patch, so that DISASSEMBLE output is much nicer.
3655 * The code in the SB-PROFILE package now seems reasonably stable.
3656 I still haven't decided what the final interface should look like
3657 (I'd like PROFILE to interact cleanly with TRACE, since both
3658 facilities use function encapsulation) but if you have a need
3659 for profiling now, you can probably use it successfully with
3660 the current CMU-CL-style interface.
3661 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3662 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3663 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3664 link now returns the dangling link itself, and for similar
3665 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3666 link itself. (In these cases the old code signalled an error and
3667 looped endlessly, respectively.) Thus, DIRECTORY now works even
3668 in the presence of dangling and cyclic symbolic links.
3669 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3670 is now a supported extension again, since the consensus on
3671 sbcl-devel was that it can be useful for ordinary development
3672 work, not just for debugging SBCL itself.
3673 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3674 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3675 that currently-defined functions might be redefined later with
3676 different return types.
3677 * Hash tables can be printed readably, as inspired by CMU CL code
3678 of Eric Marsden and SBCL code of Martin Atzmueller.
3679 * better error handling in CLOS method combination, thanks to
3680 Martin Atzmueller porting Pierre Mai's CMU CL patches
3681 * more overflow fixes for >16Mbyte I/O buffers
3682 * A bug in READ has been fixed, so that now a single Ctrl-D
3683 character suffices to cause end-of-file on character streams.
3684 In particular, now you only need one Ctrl-D at the command
3685 line (not two) to exit SBCL.
3686 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3688 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3689 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3690 does the right thing.
3691 * The implementation of some type tests, especially for CONDITION
3692 types, is now tidier and maybe faster, due to CMU CL code
3693 originally by Douglas Crosher, ported by Martin Atzmueller.
3694 * Some math functions have been fixed, and there are new
3695 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3696 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3697 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3698 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3699 likely to be built into sbcl-0.7.0.)
3700 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3701 CL, ported to SBCL by Martin Atzmueller)
3702 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3703 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3704 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3705 * There's a new slam.sh hack to shorten the edit/compile/debug
3706 cycle for low-level changes to SBCL itself, and a new
3707 :SB-AFTER-XC-CORE target feature to control the generation of
3708 the after-xc.core file needed by slam.sh.
3709 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3710 COMPILE-FILE is no longer supported, so that now every function
3711 gets an entry point, so that block compilation looks a little
3712 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3713 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3714 returns the number of bytes consed since the system started,
3715 rather than the number consed since the first time the function
3716 was called. (The new definition parallels ANSI functions like
3717 CL:GET-INTERNAL-RUN-TIME.)
3718 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3719 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3720 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3721 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3722 * Fasl file version numbers are now independent of the target CPU,
3723 since historically most system changes which required version
3724 number changes have affected all CPUs equally. Similarly,
3725 the byte fasl file version is now equal to the ordinary
3728 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3729 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3730 is no longer recognized. I apologize for this, because it was
3731 listed in SB-EXT as a supported extension, but I found that
3732 its existing behavior was poorly specified, as well as incorrectly
3733 specified, and it looked like too much of a mess to straighten it
3734 out. I have enough on my hands trying to get ANSI stuff to work..
3735 * many patches ported from CMU CL by Martin Atzmueller, with
3736 half a dozen bug fixes in pretty-printing and the debugger, and
3737 half a dozen others elsewhere
3738 * fixed bug 13: Floating point infinities are now supported again.
3739 They might still be a little bit flaky, but thanks to bug reports
3740 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3741 as flaky as they were.
3742 * The --noprogrammer command line option is now supported. (Its
3743 behavior is slightly different in detail from what the old man
3744 page claimed it would do, but it's still appropriate under the
3745 same circumstances that the man page talks about.)
3746 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3747 are now supported, and enabled by default. Thus, the compiler can
3748 handle many floating point and complex operations much less
3749 inefficiently. (Thus e.g. you can implement a complex FFT
3751 * The compiler now detects type mismatches between DECLAIM FTYPE
3752 and DEFUN better, and implements CHECK-TYPE more correctly, and
3753 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3754 thanks to patches from Martin Atzmueller.
3755 * various fixes to make the cross-compiler more portable to
3756 ANSI-conforming-but-different cross-compilation hosts (notably
3757 Lispworks for Windows, following bug reports from Arthur Lemmens)
3758 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3759 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3760 fixed thanks to Pierre Mai's CMU CL patches.
3761 * fixes to keep the system from overflowing internal counters when
3762 it tries to use i/o buffers larger than 16M bytes
3763 * fixed bug 45a: Various internal functions required to support
3764 complex special functions have been merged from CMU CL sources.
3765 (When I was first setting up SBCL, I misunderstood a compile-time
3766 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3767 * improved support for type intersection and union, fixing bug 12
3768 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3769 more obscure bugs as well
3770 * some steps toward byte-compiling non-performance-critical
3771 parts of the system, courtesy of patches from Martin Atzmueller
3772 * Christophe Rhodes has made some debian packages of sbcl at
3773 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3774 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3775 stable, but are nonetheless usable. When he's ready, I'd be happy
3776 to add them to the SourceForge "File Releases" section. (And if
3777 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3778 * new fasl file format version number (because of changes in
3779 internal representation of (OR ..) types to accommodate the new
3780 support for (AND ..) types, among other things)
3782 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3783 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3785 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3787 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3788 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3789 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3790 * DESCRIBE now works on structure objects again.
3791 * Most function call argument type mismatches are now handled as
3792 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3793 know whether the function will be redefined before the call is
3794 executed. (The compiler could flag local calls with full WARNINGs,
3795 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3796 it doesn't keep track of enough information to know whether calls
3797 are local in this sense.)
3798 * Compiler output is now more verbose, with messages truncated
3799 later than before. (There should be some supported way for users
3800 to override the default verbosity, but I haven't decided how to
3801 provide it yet, so this behavior is still controlled by the internal
3802 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3803 src/compiler/ir1util.lisp.)
3804 * Fasl file format version numbers have increased again, because
3805 support for the Gray streams extension changes the layout of the
3806 system's STREAM objects.
3807 * The Gray subclassable streams extension now works, thanks to a
3808 patch from Martin Atzmueller.
3809 * The full LOAD-FOREIGN extension (not just the primitive
3810 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3811 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3812 but like most other programs, it defaults to copying the Unix
3813 environment from the original process instead of starting the
3814 new process in an empty environment.
3815 * Extensions which manipulate the Unix environment now support
3816 an :ENVIRONMENT keyword option which doesn't smash case or
3817 do other bad things. The CMU-CL-style :ENV option is retained
3818 for porting convenience.
3819 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3820 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3822 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3824 * A patch from Martin Atzmueller seems to have solved the SIGINT
3825 problem, and as far as we know, signal-handling now works cleanly.
3826 (If you find any new bugs, please report them!)
3827 * The system no longer defaults Lisp source file names to types
3828 ".l", ".cl", or ".lsp", but only to ".lisp".
3829 * The compiler no longer uses special default file extensions for
3830 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3831 seems to expect a single default extension for all compiled code,
3832 and there's no compelling reason to try to stretch the standard
3833 to allow two different extensions.) Instead, byte-compiled files
3834 default to the same extension as native-compiled files.
3835 * Fasl file format version numbers have increased again, because
3836 a rearrangement of internal implementation packages made some
3837 dumped symbols in old fasl files unreadable in new cores.
3838 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3839 many fewer weird special cases.
3840 * Bug #17 (differing COMPILE-FILE behavior between logical and
3841 physical pathnames) has been fixed, and some related misbehavior too,
3842 thanks to a patch from Martin Atzmueller.
3843 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3844 by Tim Moore, ported to SBCL by Martin Atzmueller.
3845 * Martin Atzmueller fixed several filesystem-related problems,
3846 including bug #36, in part by porting CMU CL patches, which were
3847 written in part by Paul Werkowski.
3848 * More compiler warnings in src/runtime/ are gone, thanks to
3849 more patches from Martin Atzmueller.
3850 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3853 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3855 * DESCRIBE now works on CONDITION objects.
3856 * The debugger now handles errors which arise when trying to print
3857 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3859 * The build system now uses an additional file, customize-target-features.lisp,
3860 to allow local modifications to the target *FEATURES* list. (The point of
3861 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3862 debugging features enabled, without having to worry about propagating it
3863 into everyone's system when I do a "cvs update".) When no
3864 customize-target-features.lisp file exists, the target *FEATURES* list
3865 should be constructed the same way as before.
3866 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3867 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3868 warn about behavior which is undefined under the ANSI spec. Note: This
3869 is specified by ANSI, but it's not very popular with programmers.
3870 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3871 macro in the SBCL sources for an example of a workaround which you
3872 might use to make portable ANSI-standard code which does what you want.
3873 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3874 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3875 when it's used as a non-toplevel form. (This is required in order
3876 to implement the DEFCONSTANT-EQX macro.)
3877 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3878 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3879 funny things, due to bugs in the implementation of EVAL-WHEN
3880 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3881 non-toplevel DEFCONSTANTs are uncommon.)
3882 * The core file version number and fasl file version number have been
3883 incremented, because the old noncompliant DEFCONSTANT behavior involved
3884 calling functions which no longer exist, and because I also took the
3885 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3886 * fixed bug 1 (error handling before read-eval-print loop starts), and
3887 redid debugger restarts and related debugger commands somewhat while
3889 ** The QUIT debugger command is gone, since it did something
3890 rather different than the SB-EXT:QUIT command, and since it never
3891 worked properly outside the main toplevel read/eval/print loop.
3892 Invoking the new TOPLEVEL restart provides the same functionality.
3893 ** The GO debugger command is also gone, since you can just invoke
3894 the CONTINUE restart directly instead.
3895 ** The TOP debugger command is also gone, since it's redundant with the
3896 FRAME 0 command, and since it interfered with abbreviations for the
3898 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3900 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3901 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3902 some porting work by Martin Atzmueller)
3903 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3904 with a boatload of patches which he ported to SBCL. Now that those
3906 ** The system tries to make sure that its low-priority messages
3907 are prefixed by semicolons, to help people who like to use
3908 syntax highlighting in their ILISP buffer. (This patch
3909 was originally due to Raymond Toy.)
3910 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3911 patches originally written by Raymond Toy.
3912 ** The compiler understands coercion between single-value and
3913 multiple-VALUES type expressions better, getting rid of some very
3914 weird behavior, thanks to patches originally by Robert MacLachlan
3915 and Douglas Crosher.
3916 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3917 lambda lists, thanks to a patch originally by Pierre Mai.
3918 ** The system no longer bogusly warns about "abbreviated type
3920 ** The compiler gets less confused by inlining and RETURN-FROM,
3921 thanks to some patches originally by Tim Moore.
3922 ** The system no longer hangs when dumping circular lists to fasl
3923 files, thanks to a patch originally from Douglas Crosher.
3924 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3925 undefined function error.
3926 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3927 in structure slot initforms) for now, documented workaround instead:-|
3928 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3929 * fixed bug 5: added stubs for various Gray stream functions called
3930 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3931 installed, at least appropriate type errors are generated
3932 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3933 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3934 consistently in DEFMETHOD forms.
3935 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3936 it doesn't seem to affect SBCL after all
3937 * The C runtime system now builds with better optimization and many
3938 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3940 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3942 * The system is now under CVS at SourceForge (instead of the
3943 CVS repository on my home machine).
3944 * The new signal handling code has been tweaked to treat register
3945 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3946 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3947 downstream, e.g. in debug-int.lisp, has implicit dependencies
3948 on the unsignedness of integer representation of machine words,
3949 and that caused the system to bomb out with infinite regress
3950 when trying to recover from type errors involving signed values,
3951 e.g. (BUTLAST '(1 2 3) -1).)
3952 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3953 (This was one of the bugs Peter Van Eynde reported back in July.)
3954 * The system now uses code inspired by Colin Walters' O(N)
3955 implementation of MAP (from the cmucl-imp@cons.org mailing
3956 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3957 inline the MAP operation, and there is more than one
3958 sequence argument to the MAP call (so that it can't just
3959 do ETYPECASE once and for all based on the type of the
3960 single sequence argument). (The old non-inline implementation
3961 of the general M-argument sequence-of-length-N case required
3962 O(M*N*N) time when any of the sequence arguments were LISTs.)
3963 * The QUIT :UNIX-CODE keyword argument has been renamed to
3964 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3965 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3967 * Raymond Wiker's patches to port dynamic loading from Linux to
3968 FreeBSD have been added.
3969 * The BUGS file is now more nearly up to date, thanks in large part
3970 to Martin Atzmueller's review of it.
3971 * The debugger now flushes standard output streams before it begins
3972 its output ("debugger invoked" and so forth).
3973 * The core version number and fasl file version number have both
3974 been incremented, because of incompatible changes in the layout
3976 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3977 used to not be called for a saved Lisp image.)
3978 * Martin Atzmueller's version of a patch to fix a compiler crash,
3979 as posted on sbcl-devel 13 September 2000, has been installed.
3980 * Instead of installing Martin Atzmueller's patch for the
3981 compiler transform for SUBSEQ, I deleted the compiler transform,
3982 and transforms for some similar consing operations.
3983 * A bug in signal handling which kept TRACE from working on OpenBSD
3985 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3986 away by constant folding
3987 * The system now defines its address space constants in one place
3988 (in the Lisp sources), and propagates them automatically elsewhere
3989 (through GENESIS and the sbcl.h file). Therefore, patching the
3990 address map is less unnecessarily tedious and error-prone. The
3991 Lisp names of address space constants have also been systematized.
3992 * CVS tags like dollar-Header-dollar have been removed from
3993 the sources, because they have never saved me trouble and
3994 they've been source of trouble working with patches and other
3995 diff-related operations.
3996 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3997 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3999 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4001 * The system has been ported to OpenBSD.
4002 * The system now compiles with a simple "sh make.sh" on the systems
4003 that it's supported on. I.e., now you no longer need to tweak
4004 text in the target-features.lisp-expr and symlinks in src/runtime/
4005 by hand, the make.sh takes care of it for you.
4006 * The system is no longer so grossly inefficient when compiling code
4007 involving vectors implemented as general (not simple) vectors (VECTOR T),
4008 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4009 which dares to use the various sequence functions on non-simple
4010 vectors, takes less of a performance hit.
4011 * There is now a primitive type predicate VECTOR-T-P
4012 to test for the (VECTOR T) type, so that e.g.
4013 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4014 can now be compiled with some semblance of efficiency. (The old code
4015 turned the type declaration into a full call to %TYPEP at runtime!)
4016 * AREF on (VECTOR T) is still not fast, since it's still compiled
4017 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4018 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4019 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4020 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4021 without making full calls to SUBTYPEP at runtime.
4022 (Some analogous efficiency issues for non-simple vectors specialized to
4023 element types other than T, or for non-simple multidimensional arrays,
4024 have not been addressed. They could almost certainly be handled the
4025 same way if anyone is motivated to do so.)
4026 * The changes in array handling break binary compatibility, so
4027 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4028 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4029 instead of (VALUES T T).
4030 * By following the instructions that Dan Barlow posted to sbcl-devel
4031 on 2 July 2000, I was able to enable primitive dynamic object
4032 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4033 functionality is not implemented (since it calls ld to resolve
4034 library references automatically, requiring RUN-PROGRAM for its
4035 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4036 try to resolve library references) is now supported.
4037 * The system now flushes the standard output streams when it terminates,
4038 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4039 them at several other probably-convenient times, e.g. in each pass of
4040 the toplevel read-eval-print loop, and after evaluating a form given
4041 as an "--eval" command-line option. (These changes were motivated by a
4042 discussion of stream flushing issues on cmucl-imp in August 2000.)
4043 * The source transform for TYPEP of array types no longer assumes
4044 that an array whose element type is a not-yet-defined type
4045 is implemented as an array of T, but instead punts, so that the
4046 type will be interpreted at runtime.
4047 * There is now some support for cross-compiling in make.sh: each of
4048 the phases of make.sh has its own script. (This should be transparent
4049 to people doing ordinary, non-cross-compile builds.)
4050 * Since my laptop doesn't have hundreds of megabytes of memory like
4051 my desktop machine, I became more motivated to do some items on
4052 my to-do list in order to reduce the size of the system a little:
4053 ** Arrange for various needed-only-at-cold-init things to be
4054 uninterned after cold init. To support this, those things have
4055 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4056 symbols with such names are now uninterned after cold init).
4057 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4058 fasl files for cold load.
4059 ** Remove the old compiler structure pooling code (which used to
4060 be conditional on the target feature :SB-ALLOC) completely.
4061 ** Redo the representation of some data in cold init to be more compact.
4062 (I also looked into supporting byte compiled code at bootstrap time,
4063 which would probably reduce the size of the system a lot, but that
4064 looked too complicated, so I punted for now.)
4065 * The maximum signal nesting depth in the src/runtime/ support code has
4066 been reduced from 4096 to 256. (I don't know any reason for the very
4067 large old value. If the new smaller value turns out to break something,
4068 I'll probably just bump it back up.)
4069 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4071 * Many, many bugs reported by Peter Van Eynde have been added to
4072 the BUGS list; some have even been fixed.
4073 * While enabling dynamic object file loading, I tried to make the
4074 code easier to understand, renaming various functions and variables
4075 with less ambiguous names, and changing some function calling
4076 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4077 * While trying to figure out how to do the OpenBSD port, I tried to
4078 clean up some of the code in src/runtime/. In particular, I dropped
4079 support for non-POSIX signal handling, added various comments,
4080 tweaked the code to reduce the number of compilation warnings, and
4081 renamed some files to increase consistency.
4082 * To support the new automatic configuration functionality in make.sh,
4083 the source file target-features.lisp-expr has been replaced with the
4084 source file base-target-features.lisp-expr and the machine-generated
4085 file local-target-features.lisp-expr.
4086 * fixed a stupid quoting error in make.sh so that using CMU CL
4087 "lisp -batch" as cross-compilation host works again
4089 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4091 * DESCRIBE no longer tries to call itself recursively to describe
4092 bound/fbound values, so that it no longer fails on symbols which are
4093 bound to themselves (like keywords, T, and NIL).
4094 * DESCRIBE now works on generic functions.
4095 * The printer now prints less-screwed-up representations of closures
4096 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4097 * A private symbol is used instead of the :EMPTY keyword previously
4098 used to mark empty slots in hash tables. Thus
4099 (DEFVAR *HT* (MAKE-HASH-TABLE))
4100 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4101 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4102 now does what ANSI says that it should. (You can still get
4103 similar noncompliant behavior if bang on the hash table
4104 implementation with all the symbols you get back from
4105 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4106 This breaks binary compatibility, since tests for equality to
4107 :EMPTY are wired into things like the macroexpansion of
4108 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4110 * There's now a minimal placeholder implementation for CL:STEP,
4111 as required by ANSI.
4112 * An obscure bug in the interaction of the normal compiler, the byte
4113 compiler, inlining, and structure predicates has been patched
4114 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4115 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4117 * Missing ordinary arguments in a macro call are now detected even
4118 when the macro lambda list contains &KEY or &REST.
4119 * The debugger no longer complains about encountering the top of the
4120 stack when you type "FRAME 0" to explicitly instruct it to go to
4121 the top of the stack. And it now prints the frame you request even
4122 if it's the current frame (instead of saying "You are here.").
4123 * As specified by ANSI, the system now always prints keywords
4124 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4126 * The default initial SIZE of HASH-TABLEs is now smaller.
4127 * Type information from CLOS class dispatch is now propagated
4128 into DEFMETHOD bodies, so that e.g.
4129 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4131 is now basically equivalent to
4132 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4133 (DECLARE (TYPE SINGLE-FLOAT X))
4135 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4136 operation, without having to do run-time type dispatch.
4137 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4138 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4139 * Since I don't seem to be making big file reorganizations very often
4140 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4141 is overflowing my ability to conveniently back them up), I've finally
4142 checked the system into CVS. (The CVS repository is on my home system,
4143 not at SourceForge -- putting it on SourceForge might come later.)
4144 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4145 high-level GC-NOTIFY-FOO functions send their output. (There's
4146 still very little control of where low-level verbose GC functions
4147 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4148 less than it used to -- the GC-NOTIFY-FOO functions are now under
4149 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4150 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4151 in only one place in the source code, and propagates it automatically
4152 everywhere that it's needed. Thus e.g. when I bump the version from
4153 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4154 * The C source files now include boilerplate legalese and documentation
4155 at the head of each file (just as the Lisp source files already did).
4156 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4157 to his page will be replaced with a link to his new CLiki service.
4159 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4161 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4162 * The build process now looks for GNU make under the default name "gmake",
4163 instead of "make" as it used to. If GNU make is not available as "gmake"
4164 on your system, you can change this default behavior by setting the
4165 GNUMAKE environment variable.
4166 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4167 can build without error under CMU CL.
4169 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4171 * There is now a partial SBCL user manual (with some new text and some
4172 text cribbed from the CMU CL manual).
4173 * The beginnings of a profiler have been added (starting with the
4174 CMU CL profiler and simplifying and cleaning up). Eventually the
4175 main interface should be through the TRACE macro, but for now,
4176 it's still accessed through vaguely CMU-CL-style functions and macros
4177 exported from the package SB-PROFILE.
4178 * Some problems left over from porting CMU CL to the new
4179 cross-compilation bootstrap process have been cleaned up:
4180 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4181 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4182 part when getting rid of the compiler *BACKEND* stuff.)
4183 ** The value of *NULL-TYPE* was screwed up, because it was
4184 being initialized before the type system knew the final
4185 definition of the 'NULL type. This screwed up several key
4186 optimizations in the compiler, causing inefficiency in all sorts
4187 of places. (I found it because I wanted to understand why
4188 GET-INTERNAL-RUN-TIME was consing.)
4189 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4190 PROCLAIM FTYPE information. Unfortunately this broke binary
4191 compatibility again, since now the forms output by DEFGENERIC
4192 to refer to functions which didn't exist in 0.6.3.
4193 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4194 can use the new (as of 0.6.3) transform for SOME into MAP into
4196 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4197 values to (INTEGER 0 1000000), so that the time code will no longer
4198 occasionally get blown up by Linux returning 1000000 microseconds
4199 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4200 its output conform to the ANSI spec. (Alas, this makes its output
4201 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4202 * A full call to MAP NIL with a single sequence argument no longer conses.
4203 * fixes to problems pointed out by Martin Atzmueller:
4204 * The manual page no longer talks about multiprocessing as though
4205 it were currently supported.
4206 * The ILISP support patches have been removed from the distribution,
4207 because as of version 5.10.1, ILISP now supports SBCL without us
4208 having to maintain patches.
4209 * added a modified version of Raymond Toy's recent CMU CL patch for
4210 EQUALP comparison of HASH-TABLE
4212 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4214 * The system still can't cross-compile itself with
4215 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4216 entails), but at least it can compile more of itself that way
4217 than it used to be able to, and various buglets which were uncovered
4218 by trying to cross-compile itself that way have now been fixed.
4219 * This release breaks binary compatibility again. This time
4220 at least I've incremented the FASL file format version to 2, so that the
4221 problem can be detected reliably instead of just causing weird errors.
4222 * various new style warnings:
4223 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4224 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4225 ** using the lexical binding of a variable named in the *FOO* style
4226 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4227 as specified by ANSI.
4228 * *RANDOM-STATE* is no longer automatically initialized from
4229 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4230 default behavior of the system is to repeat its behavior every time
4231 it's run. If you'd like to change this behavior, you can always
4232 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4233 old convention there was no comparably easy way to get the system to
4234 repeat its behavior every time it was run.
4235 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4236 FTYPE declarations has been removed, in favor of their ANSI
4237 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4238 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4239 the types of their sequence arguments can be determined at compile time.
4240 This is done through a new open code expansion for MAP which eliminates
4241 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4242 argument types can be determined at compile time.
4243 * The optimizer now transforms COERCE into an identity operation when it
4244 can prove that the coerced object is already of the correct type. (This
4245 can be a win for machine generated code, including the output of other
4246 optimization transforms, such as the MAP transform above.)
4247 * Credit information has been moved from source file headers into CREDITS.
4248 * Source file headers have been made more standard.
4249 * The CASE macro now compiles without complaining even when it has
4252 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4254 * (Note that the way that the PCL macroexpansions were rewritten
4255 to accommodate the change in DEFGENERIC below breaks binary
4256 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4257 not run under sbcl-0.6.2. Once we get out of alpha releases,
4258 i.e. hit release 1.0.0, we'll probably try to maintain binary
4259 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4260 and sbcl-1.4.4. Until then, however, it might be fairly common
4261 for maintenance releases to break binary compatibility.)
4262 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4264 * The reporting of unhandled signals has been changed to print some
4265 explanatory text as well as the report form. (Previously only
4266 the report form was printed.)
4267 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4268 it defines, so that the compiler no longer issues undefined function
4269 warnings for compiled-but-not-yet-loaded generic functions.
4270 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4271 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4272 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4273 section of the ANSI spec are supported.
4274 * The "" nickname for the "KEYWORD" package has been removed.
4275 The reader still handles symbol tokens which begin with a package marker
4276 as keywords, but it doesn't expose its mechanism for doing so in the
4277 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4278 * The system now issues STYLE-WARNINGs for contradictory TYPE
4279 proclamations. (Warnings for contradictory FTYPE proclamations would
4280 be nice too, but those can't be done usefully unless the type system
4281 is made smarter about FUNCTION types.)
4282 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4283 systematized, so that "*target-*.lisp is supposed to exist only on the
4284 target and imply that there's a related file which exists on the
4285 host, and *host-*.lisp is supposed to exist only on the host and imply
4286 that there's a related file which exists on the target. This involves a
4287 lot of renaming. Hopefully the acute confusion caused by the renaming
4288 will be justified by the reduction in chronic confusion..
4289 ** runtime-type.lisp -> early-target-type.lisp
4290 ** target-type.lisp -> late-target-type.lisp
4291 ** early-host-format.lisp -> early-format.lisp
4292 ** late-host-format.lisp -> late-format.lisp
4293 ** host-error.lisp -> misc-error.lisp
4294 ** early-error.lisp -> early-target-error.lisp
4295 ** late-error.lisp -> late-target-error.lisp
4296 ** host-defboot.lisp -> early-defboot.lisp
4297 ** code/misc.lisp -> code/target-misc.lisp
4298 ** code/host-misc.lisp -> code/misc.lisp
4299 ** code/numbers.lisp -> code/target-numbers.lisp
4300 ** code/early-numbers.lisp -> numbers.lisp
4301 ** early-host-type.lisp -> early-type.lisp
4302 ** late-host-type.lisp -> late-type.lisp
4303 ** host-typep.lisp -> typep.lisp
4304 ** load.lisp -> target-load.lisp
4305 ** host-load.lisp -> load.lisp
4306 ** host-disassem.lisp -> disassem.lisp
4307 ** host-insts.lisp -> insts.lisp
4308 ** byte-comp.lisp -> target-byte-comp.lisp
4309 ** host-byte-comp.lisp -> byte-comp.lisp
4310 ** host-signal.lisp -> signal.lisp
4311 ** host-defstruct.lisp -> defstruct.lisp
4312 ** late-target-type.lisp -> deftypes-for-target.lisp
4313 Furthermore, several other previously target-only files foo.lisp (e.g.
4314 hash-table.lisp and random.lisp) have been split into a target-and-host
4315 foo.lisp file and a target-only target-foo.lisp file, with their key type
4316 definitions in the target-and-host part, so that the cross-compiler will
4317 know more about target types.
4318 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4319 gone away. In their place are various *BACKEND-FOO* variables
4320 corresponding to the slots of the old structure.
4321 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4322 into the target SBCL has been fixed.
4323 * The chill.lisp system for loading cold code into a running SBCL
4325 * Support for the CMU CL "scavenger hook" extension has been removed.
4326 (It was undocumented and unused in the CMU CL sources that SBCL was
4327 derived from, and stale in sbcl-0.6.1.)
4328 * Various errors in the cross-compiler type system were detected
4329 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4330 (enabling various consistency checks). Many of them were fixed,
4331 but some hard problems remain, so the compiler is back to
4332 running without *TYPE-SYSTEM-INITIALIZED* for now.
4333 * As part of the cross-compiler type system cleanup, I implemented
4334 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4335 * I have started adding UNCROSS calls throughout the type system
4336 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4337 on cross-compiler input files will be able to go away, and various
4339 * CONSTANTP now returns true for quoted forms (as explicitly required
4342 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4344 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4345 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4346 (MAKE-LIST -1). In the longer term, it ought to become true
4347 that declarations are assertions even at SAFETY 1. For now, it's not
4348 quite true even at SAFETY 3, but it's at least more nearly true..
4349 (Note that this change seems to increases the size of the system by
4350 O(5%) and to decrease the speed of the compiler by 20% or more.)
4351 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4352 for the problem of printing dozens of lines of distracting information
4353 about low-level system machinery as part of the top stack frame
4354 on entry to the debugger when an undefined function was called.
4355 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4356 is not reset to COMMON-LISP-USER.
4357 * Compilation of stuff related to dyncount.lisp has been made conditional
4358 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4359 smaller. The various dyncount-related symbols have been moved into
4360 a new "SB-DYNCOUNT" package.
4361 * tty-inspect.lisp has been renamed to inspect.lisp.
4362 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4363 feature has gone away. (When we eventually port to other flavors of
4364 libc and/or Unix, we'll try to make the differences between flavors
4365 invisible at the user level.)
4366 * Various other *FEATURES* tags, and/or their associated conditionals,
4367 have been removed if obsolescent, or given better documentation, or
4368 sometimes given more-mnemonic names.
4370 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4372 * tidied up "make.sh" script
4373 * tidied up system directory structure
4374 * better "clean.sh" behavior
4375 * added doc/FOR-CMUCL-DEVELOPERS
4376 * many many small tweaks to output format, e.g. removing possibly-confusing
4377 trailing #\. character in DESCRIBE-INSTANCE
4378 * (EQUALP #\A 'A) no longer signals an error.
4379 * new hashing code, including EQUALP hashing
4380 * tidied up Lisp initialization and toplevel
4381 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4382 * command line argument processing
4383 * added POSIX-GETENV function to deal with Unix-ish environment variables
4384 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4385 terminating SBCL on EOF
4386 * non-verbose GC by default
4387 * There is no more "sbcl" shell script; the sbcl file is now the C
4388 runtime executable (just like CMU CL).
4389 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4391 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4392 symbol and package information as Lisp native symbol and package objects
4393 instead of strings naming symbols and strings naming packages. This way,
4394 whenever packages are renamed (as in warm init), debug information is
4395 transformed along with everything else.
4396 * tweaked the optimization policy declarations which control the building
4397 of SBCL itself. Now, among other things, the system no longer saves
4398 source location debugging information. (This helps two problems at once
4399 by reducing SBCL size and by keeping SBCL from trying to look for its
4400 sources -- which may not exist -- when reporting errors.)
4401 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4402 debugging and testing purposes
4403 * cleaned up printing, making the printer call PRINT-OBJECT for
4404 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4405 methods, giving nearly-ANSI behavior
4406 * converted almost all special variables to use *FOO* naming convention
4407 * deleted PARSE-TIME functionality, since it can be done portably
4408 * moved some files out of cold init into warm init
4409 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4411 * regularized formatting of source files
4412 * added an install.sh script
4413 * fixed ridiculous memory usage of cross-compiler by making
4414 compiler/alloc.lisp not try to do pooling unless it can hook
4415 itself into the GC of the cross-compilation host. Now the system
4416 builds nicely on my old laptop.
4417 * added :SB-ALLOC in target-features.lisp-expr
4418 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4419 was not implemented)
4420 * re-did condition handling and note reporting in the compiler. Notes
4421 are no longer handled by signalling conditions. Style warnings
4422 and warnings are handled more correctly and reported in such a way
4423 that it's easy to find one or the other in your output (so that you
4424 can e.g. figure out which of many problems caused COMPILE-FILE to
4426 * changed the severity of several compiler warnings from full WARNING
4427 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4428 compiler note reporting so that it doesn't use the condition system
4429 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4430 COMPILE-FILE command)
4431 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4432 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4433 no longer supported, since their implementation was deeply intertwingled
4434 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4435 * removed bogus "support" for compiler macros named (SETF FOO), and
4436 removed the compiler macro for SETF INFO (but only after making a fool
4437 of myself on the cmucl-imp mailing list by posting a bogus patch for
4438 DEFINE-COMPILER-MACRO..)
4439 * Compiled files containing forms which have side effects on the Lisp
4440 reader (such as DEFPACKAGE forms) are now handled more correctly.
4441 (Compiler queuing of top level lambdas has been suppressed by setting
4442 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4443 * deleted various currently-unused source files, e.g. gengc.lisp. They
4444 may be added back at some point e.g. when porting to other architectures,
4445 but until they are it's distracting to distribute them and to try to
4447 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4448 there were so many of them they're just distractions, and UNCROSS is
4449 known to be able to handle the current sources
4450 * moved PROFILE functionality into TRACE, so that it will be clear
4451 how the wrapping and unwrapping of functions when you profile them
4452 interacts with the wrapping and unwrapping of functions when you
4453 trace them. (Actually, the functionality isn't there yet, but at least
4454 the interface specification is there. Hopefully, the functionality will
4455 arrive with some future maintenance release.)
4456 * removed host-oops.lisp
4457 * changed signature of QUIT function to allow UNIX-CODE argument
4458 * fixed READ-SEQUENCE bug
4459 * tweaked verbose GC output so that it looks more like the progress
4460 output that ANSI specifies for functions like LOAD
4461 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4462 * added <http://sbcl.sourceforge.com> to the banner information printed by