1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.17 relative to 1.0.16:
3 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
5 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
6 * optimization: two argument forms of LAST are upto ~10% faster
7 in normal SPEED policies.
8 * optimization: NCONC no longer needs to heap cons its &REST list
9 in normal SPEED policies.
10 * bug fix: LAST when always returned the whole list when given a bignum
11 as the second argument.
12 * bug fix: dynamic extent allocation of nested lists and vectors
13 could leak to otherwise accessible parts.
14 * bug fix: invalid optimization of heap-allocated alien variable
16 * bug fix: fasl header checking is less vulnerable to different
17 platform word lengths.
18 * bug fix: more correct assembler syntax for GNU binutils
19 2.18.50.0.4 support. (thanks to Marijn Schouten)
21 changes in sbcl-1.0.16 relative to 1.0.15:
22 * minor incompatible change: revert the changes to sb-posix's error
23 signaling added in 1.0.14.
24 * minor incompatible change: change PROBE-FILE back to returning
25 NIL whenever we can't get a truename, as was the case before 1.0.14.
26 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
27 inherits from ERROR instead of just CONDITION.
28 * new feature: SB-INTROSPECT can provide source locations for instances
29 as well. (thanks to Tobian Ritterweiler)
30 * optimization: binding special variables now generates smaller code
31 on threaded platforms.
32 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
33 and cases where no :TEST is given but the compiler can infer that
34 the element to search is of type (OR FIXNUM (NOT NUMBER)).
35 * optimization: better LOGNOT on fixnums.
36 * optimization: modular arithmetic for a particular requested width
37 is implemented using a tagged representation unless a better
38 representation is available.
39 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
40 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
41 for wellformedness and returns a specialized vector. (reported by
43 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
44 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
45 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
46 unscheduling timer at the same time as another timer fires.
47 * bug fix: don't reschedule timers for dead threads.
48 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
49 * bug fix: copying output from RUN-PROGRAM to a stream signalled
50 bogus errors if select() was interrupted.
51 * enhancement: add support for fcntl's struct flock to SB-POSIX.
53 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
54 * enhancement: cleaner backtraces for interactive interrupts, as
55 well as other cases where the interesting frames used to be
56 obscured by interrupt handling frames.
57 * enhancement: untracing a whole package using (UNTRACE "FOO") is
58 now supported, and tracing a whole package using (TRACE "FOO") now
59 traces SETF-functions as well.
60 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
61 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
62 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
63 * unreadably printed representation of hash-tables now includes
65 * bug fix: partially fixed #188: type propagation from assignments
66 is now more efficient.
67 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
68 'DOUBLE-FLOAT) are not flushable.
69 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
70 full address of the object, and none of the tag bits.
71 * bug fix: readably printing hash-tables now respects other printer
72 control variables. (reported by Cedric St-Jean)
73 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
74 keyword with MAKE-HASH-TABLE.
75 * bug fix: export SB-POSIX:MKSTEMP.
76 * bug fix: SORT was not interrupt safe.
77 * bug fix: XREF accounts for the last node of each basic-block as
79 * bug fix: MAKE-INSTANCE optimizations interacted badly with
80 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
81 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
82 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
83 create obsolete instances in the case of redefinition or
84 obsoletion of a superclass. (thanks to Andy Hefner)
85 * bug fix: Support for the Alpha architecture has been revived; it had
86 suffered somewhat from lack of maintenance since sbcl-1.0.
87 * improvements to the Windows port:
88 ** The system detects the codepage to use at startup, instead of
89 using the value from the saved core. (thanks to Kei Suzuki)
91 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
92 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
93 (see documentation for details.)
94 * revived support for OpenBSD (contributed by Josh Elsasser)
95 * partially fixed bug #108: ROOM no longer suffers from occasional
96 (AVER (SAP= CURRENT END)) failures.
97 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
98 bodies no longer cause a WARNING to be signalled. (reported by
100 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
101 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
102 the bug in TRUNCATE.) (reported by Michael Weber)
103 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
104 no samples. (reported by Andy Hefner)
105 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
106 no longer appear as (NIL ...) frames in backtraces.
107 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
108 is now actually exported from SB-EXT as documented. (reported by
110 * bug fix: sb-aclrepl now correctly understands how to inspect
111 single-floats on 64-bit platforms where single-floats are not boxed.
112 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
113 yet been finalized. (reported by Levente Meszaros)
114 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
115 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
116 SBCL-specific optimize qualities.
118 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
119 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
120 an executable in the search path, and does so in the child
121 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
122 been removed; it can be found in the CVS history, for somebody who
123 needs that search behavior (see the manual).
124 * minor incompatible change: compiler policy re. weakening type
125 checks has changed: now type checks are weakened only if SAFETY < 2
127 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
128 unparsing of directory pathnames as files. Analogously,
129 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
130 filename to parse into a directory pathname.
131 * enhancement: implicit generic function creation now signals a
132 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
133 users can bind handlers for and muffle around calls to LOAD.
134 * enhancement: RUN-PROGRAM allows unicode arguments and environments
135 to be used (using the default stream external format), and allows
136 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
137 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
138 strings and vectors whose element-type or simplicity is not fully
139 known at compile-time.
140 * optimization: STRING-TRIM and related functions no longer allocate
141 a new string when no trimming needs to be performed. These functions
142 are also faster than before when the input string has been declared
144 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
145 * optimization: READ-LINE is significantly faster for files containing
147 * optimization: non-open coded uses of character comparison operators
148 (e.g. char=) no longer cons when called with more than one parameter
149 on platforms supporting dynamic-extent allocation.
150 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
151 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
152 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
154 * bug fix: some sequence functions elided bounds checking when
156 * bug fix: too liberal weakening of union-type checks when SPEED >
158 * bug fix: more bogus fixnum declarations in ROOM implementation
161 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
162 * new feature: MAKE-HASH-TABLE now experimentally accepts a
163 :SYNCHRONIZED argument, which makes the hash-table safe for
164 concurrent accesses (but not iteration.) See also:
165 SB-EXT:WITH-LOCKED-HASH-TABLE, and
166 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
167 * optimization: CONCATENATE on strings is an order of magnitude faster
168 in code compiled with (> SPEED SPACE).
169 * optimization: SUBSEQ is ~50% faster on lists.
170 * bug fix: bug 417 fixed -- source location reporting is now more robust.
171 * bug fix: SUBSEQ on a list will now correctly signal an error if if
172 END is smaller then START.
173 * bug fix: SB-PROFILE will no longer report extra consing for nested
174 calls to profiled functions.
175 * bug fix: ROOM implementation had bogus fixnum declarations which
176 could cause type-errors when calling ROOM on large images.
177 * bug fix: if file compilation is aborted, the partial fasl is now
178 deleted, and COMPILE-FILE returns NIL as the primary value.
179 * bug fix: number of thread safety issues relating to SBCL's internal
180 hash-table usage have been fixed.
181 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
182 be returned from its body when the values were being returned
183 using unknown-values return convection and the W-P-O was wrapped
184 inside an UNWIND-PROTECT.
185 * bug fix: sb-posix should now compile again under Windows, enabling
188 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
189 * incompatible change: hash-table accessor functions are no longer
190 automatically protected by locks. Concurrent accesses on the same hash-table
191 from multiple threads can give inconsistent results or even corrupt the
192 hash-table completely. Multi-threaded applications should do their own
193 locking at the correct granularity. In the current implementation it is
194 still safe to have multiple readers access the same table, but it's not
195 guaranteed that this property will be maintained in future releases.
196 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
197 and will signal an error at runtime.
198 * enhancement: SB-THREAD package now exports a semaphore interface.
199 * enhancement: CONS can now stack-allocate on x86 and
200 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
202 * enhancement: nested lists can now be stack allocated on
203 platforms providing stack allocation support.
204 * enhancement: dynamic-extent support has been extended to support
205 cases where there are multiple possible sources for the stack
207 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
208 if the mutex is uncontested on Linux.
209 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
210 as specified by the CLHS. (thanks to Tobias Rittweiler)
211 * bug fix: NaN comparison now works on x86-64.
212 * bug fix: CLOSE :ABORT T on a stream with pending output now
214 * bug fix: instances of non-standard metaclasses using standard
215 instance structure protocol sometimes missed the slot type checks
217 * bug fix: known functions can be inlined in high-debug code.
219 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
220 * minor incompatible change: the MSI installer on Windows no longer
221 associates .lisp and .fasl files with the installed SBCL.
222 * minor incompatible change: :UNIX is no longer present in *FEATURES*
223 on Windows. (thanks to Luis Oliviera)
224 * new feature: SB-CLTL2 contrib module now implements
225 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
226 * optimization: scavenging weak pointers is now more efficient,
227 requiring O(1) instead of O(N) per weak pointer to identify
228 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
229 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
230 method bodies using constant slot-names when the first argument is
231 a specializer parameter for the method.
232 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
233 names, when the class of the instance is a direct instance
234 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
235 as slow as the constant slot-name case.)
236 * optimization: member type construction is now O(N) instead
238 * optimization: UNION and NUNION are now O(N+M) for large
239 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
241 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
242 now more readable in environments like Slime which display it.
243 (thanks to Tobias C. Rittweiler)
244 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
245 was able to infer the correct argument type for the object on which
246 the CAS operation was being performed.
247 * bug fix: copy propagation interfered with parallel assignment
248 semantics in local calls. (reported by Paul Khuong)
249 * bug fix: the signed modular fixnum shift compiled to wrong code on
250 x86 and x86-64. (spotted by a slight modification to some of
253 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
254 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
255 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
256 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
258 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
259 and generic function now print the object identity for anonymous
260 instances. (thanks to Attila Lendvai)
261 * enhancement: as an extension to MOP, SBCL now supports SETF of
262 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
263 (thanks to Attila Lendvai)
264 * enhancement: sb-cover has an alternate annotation mode which puts
265 the form annotations at the car of the form, instead of annotating
266 it completely. The benefit of this mode is that it shows better
267 which forms were instrumented by the compiler.
268 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
269 by printer control variables. (reported by Dan Corkill)
270 * bug fix: system leaked memory when delayed output was performed by
271 the OS in smaller chunks then expected. (thanks to David Smith)
272 * bug fix: system leaked memory when file streams were not closed
274 * bug fix: large objects written to slow streams that were modified
275 after the write could end up with the modified state written to
276 the underlying file descriptor.
277 * bug fix: multiple threads operating in parallel on the same stream
278 could cause buffer-overflows.
279 * bug fix: source location information is stored correctly
280 (broken since 1.0.6). This bug would generally show up as the
281 Slime debugger highlighting the wrong form when the sldb-show-source
283 * bug fix: Branch forms are again annotated as branches in the sb-cover
285 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
286 on x86-64. (reported by Gregory Vanuxem)
287 * bug fix: EQUALP could return wrong results for structures with raw
288 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
289 word). (reported by Vjacheslav Fyodorov)
290 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
293 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
294 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
295 atomic compare-and-swap operations on threaded platforms.
296 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
297 allows assigning a global minimum value to optimization qualities
298 (overriding proclamations and declarations).
299 * enhancement: closed over variables can be stack-allocated on x86
301 * performance bug fix: GETHASH and (SETF GETHASH) are once again
303 * optimization: slot definition lookup is now O(1). This speeds up
304 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
305 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
306 * optimization: ASSOC and MEMBER can now be open-coded for all
307 combinations of keyword arguments when second argument is constant
308 and SPEED >= SPACE. In other cases a specialized version is
310 * bug fix: using obsoleted structure instances with TYPEP and
311 generic functions now signals a sensible error.
312 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
313 (reported by Kristoffer Kvello)
314 * bug fix: backtrace construction is now more careful when making
315 lisp-objects from pointers on the stack, to avoid creating bogus
316 objects that can be seen by the GC.
317 * bug fix: defaulting of values in contexts expecting more than 7
318 variables now works on x86-64. (reported by Christopher Laux)
319 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
321 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
322 as the property-list of a symbol.
323 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
324 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
325 situations requested, are once again file-compileable. (reported
328 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
329 * MOP improvement: support for user-defined subclasses of
330 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
331 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
332 the expansion of DEFMETHOD, is responsible for generating a form
333 which creates a list of specializers when evaluated. Additional
334 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
335 debugging and introspective support.
336 * minor incompatible change: the (unsupported) spinlock interface
337 has changed: free spinlock now has the value NIL, and a held spinlock
338 has the owning thread as its value.
339 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
340 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
342 * enhancement: name of a socket-stream is now "a socket" instead of
344 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
345 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
346 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
347 for generic functions. (Thanks to Tobias C. Rittweiler)
348 * optimization: bignum printing speed has been improved by 20-40%
349 (depending on the bignum size.)
350 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
352 * bug fix: the cache used by the CLOS to store precomputed effective
353 methods, slot offsets, and constant return values is now thread and
355 * bug fix: generic function dispatch function updating is now thread
356 and interrupt safe (in the sense that the known issues have been
358 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
359 * bug fix: interning EQL-specializers is now thread and interrupt safe.
360 * bug fix: asdf systems with dependencies to the SB-POSIX or
361 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
362 * bug fix: interrupt safety of applicable method computation has been
365 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
366 * new contrib: sb-cover, an experimental code coverage tool, is included
368 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
369 significantly faster.
370 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
371 produce more efficient code on x86-64 if the compiler can determine
372 that (+ INDEX <constant>) does not require a bounds check and FOO
373 has an element type at least 8 bits wide.
374 * enhancement: a new, experimental synchronous timeout facility is
375 provided. Refer to SB-SYS:WITH-DEADLINE for details.
376 * enhancement: when a symbol name conflict error arises, the
377 conflicting symbols are always printed with a package prefix.
378 (thanks to Kevin Reid)
379 * enhancement: stepping is now once again supported on the SPARC and
380 MIPS platforms. (It is also now more likely to work on CheneyGC
382 * enhancement: sb-sprof can now also track and report accurate call
384 * bug fixes: the treatment of non-standard subclasses of
385 SB-MOP:SPECIALIZER is more correct.
386 * incompatible change: PURIFY no longer copies the data from the
387 dynamic space into the static and read-only spaces on platforms
388 that use the generational garbage collector
389 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
391 * bug fix: GC race condition occasionally resulting in crashes with
392 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
394 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
396 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
397 a STYLE-WARNING to be signalled (regression from 1.0.4.)
398 * bug fix: an asynchronous interrupt could previously leave the
399 system running with GC inhibited.
400 * bug fix: a DECLARE form evaluated at top-level now causes an error
401 rather than silently (or verbosely) returning NIL.
402 * bug fix: trying to dissassemble functions compiled at high DEBUG could
403 result in a error being signalled due to source form lookup errors.
404 (reported by Peter Graves)
406 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
407 * incompatible change: removed writer methods for host-ent-name,
408 host-ent-addresses -- changing the values did not update the DNS
410 * minor incompatible change: changed experimental JOIN-THREAD interface
411 * documentation: the manual now lists reader and writer methods
412 in class slot documentation sections. (thanks to Richard M Kreuter)
413 * documentation: unwinding from asyncronous events has been
414 documented as unsafe.
415 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
416 in multithreaded application code.
417 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
418 platforms. (thanks to James Anderson for the optimization hint)
419 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
421 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
422 variants no longer cons.
423 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
424 their NOT- variants no longer cons.
425 * optimization: EQUAL hash tables no longer use SXHASH for objects
426 of all data types, but instead use an EQL hash for types for which
427 EQUAL is the same as EQL.
428 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
429 are significantly faster.
430 * optimization: new STRING-OUTPUT-STREAM implementation is much
431 faster for large outputs and conses 30% less on average.
432 * enhancement: XREF information is now collected for references made
433 to global variables using SYMBOL-VALUE and a constant argument.
434 * enhancement: SIGINT now causes a specific condition
435 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
436 * bug fix: FILE-POSITION used to signal an error in some cases where
437 ANSI requires it to return NIL.
438 * bug fix: ADJUST-ARRAY is now interrupt-safe.
439 * bug fix: adding and removing fd-handlers is now interrupt-safe.
440 * bug fix: inlined calls to C now ensure 16byte stack alignment on
442 * bug fix: bad type declaration in the CLOS implementation has
443 been fixed. (reported by James Anderson)
444 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
445 (thanks to Lutz Euler)
446 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
447 required by the CLHS. (reported by Samium Gromoff)
448 * bug fix: dead unbound variable references now signal an error.
449 (reported by Marco Monteiro)
450 * bug fix: / with an unused value was being deleted in safe code.
451 (thanks to Marco Monteiro and Kevin Reid)
452 * bug fix: number of characters that can be written onto a single
453 line in a file is unlimited.
454 * bug fix: some GC deadlocks caused by asynchronous interrupts have
455 been fixed by inhibiting interrupts for when GC is disabled.
456 * bug fix: some interrupt safety issues with GETHASH, (SETF
457 GETHASH), CLRHASH and REMHASH have been fixed.
458 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
459 specifier no longer causes infinite recursion.
460 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
461 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
462 * bug fix: an &environment argument with the correct variable information
463 is passed to macros that are expanded during byte compilation.
464 (reported by Samium Gromoff)
465 * bug fix: a logic error was causing the hash function for bignums to
466 have excessive amounts of collisions. (reported by Faré Rideau)
467 * bug fix: modifying the contents of an array could change the return
468 value of SXHASH on that array, which is only allowed for strings
469 and bit vectors. (bug introduced in 0.9.16)
470 * improvement: the x86-64/darwin port now passes all tests (except
471 for the debugger tests) but should still be considered
472 experimental until this is fixed.
473 * improvement: a style-warning is signaled for CASE (etc) clauses with
474 duplicate keys. (thanks to Kevin Reid)
475 * improvement: macos/x86 and macos/x86-64 now use mach exception
476 handlers for dealing with illegal instructions (for trapping and
477 error handling) and memory protection violations (for GC).
479 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
480 * new platform: experimental support for x86-64/darwin (MacOS).
481 * incompatible change: the thread-safe (on most platforms) getaddrinfo
482 and getnameinfo sockets functions are used instead of gethostbyaddr
483 and gethostbyname, on platforms where the newer functions are available.
484 As a result, the ALIASES field of HOST-ENT will always be NIL on these
486 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
487 don't print the startup banner, but behave as if --noinform was passed
488 as a command line argument. (thanks to Kevin Reid)
489 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
490 * optimization: code using alien values with undeclared types is much faster.
491 * optimization: the compiler is now able to open code SEARCH in more cases.
492 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
493 * bug fix: using standardized COMMON-LISP special variables as loop
494 variables no longer signals bogus package lock violations. (reported
496 * bug fix: declaring local loop variables to be of a range-limited type
497 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
498 (reported by Andras Simon)
499 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
500 bugs remain on x86-64.)
501 * bug fix: the #= and ## reader macros now interact reasonably with
502 funcallable instances.
503 * bug fix: type-checks for function arguments were compiled using the
504 compiler policy settings of the wrong lexical environment (in 1.0.2
506 * bug fix: SHADOW accepts characters as string designators, as required
507 by the spec (thanks to Eric Marsden)
508 * bug fix: fixed GC safety issues when foreign functions are called with
509 non-base strings as arguments
510 * bug fix: more consistent error output for fatal-compiler-warnings, like
512 * bug fix: functions with proper names were showing up as NIL in
513 backtraces (reported by Edi Weitz, regression in 1.0.2)
515 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
516 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
517 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
518 produce more efficient code on the x86 if the compiler can determine
519 that (+ INDEX <constant>) does not require a bounds check and FOO
520 has an element type at least 8 bits wide.
521 * bug fix: references to undefined variables are handled the same way
522 in toplevel forms as in the normal compiler
523 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
525 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
526 (thanks to Jon Buller)
527 * improvement: faster compilation times for complex functions
528 * improvement: added readlink support to SB-POSIX (thanks to Richard
531 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
532 * improvement: experimental support for mach exception handling on
533 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
535 * improvement: support for GBK external format.
536 (thanks to Chun Tian (binghe))
537 * improvement: the debugger now displays variables that have been closed
538 over, in code compiled with (DEBUG 2) or higher
539 * improvement: support for executable cores on NetBSD (thanks to
541 * new feature: added a RESTART-FRAME debugger command
542 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
543 be used to provide an implementation for FILE-POSITION on Gray streams
544 (thanks to Eric Marsden)
545 * new feature: add syslog, openlog and closelog support to SB-POSIX
546 (thanks to Richard Kreuter)
547 * optimization: the function call overhead in code compiled with
548 a high DEBUG optimization setting is significantly
549 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
550 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
551 (thanks to Tony Martinez)
552 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
553 evaluated once (reported by Kevin Reid)
554 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
555 works again (thanks to Kilian Sprotte)
556 * bug fix: an error is signaled for tagbodies with duplicate tags
557 (thanks to Stephen Wilson)
558 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
559 * bug fix: Win32 port can now handle foreign code unwinding Lisp
560 stack frames from alien callbacks.
561 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
562 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
563 65k pages (thanks to David Woodhouse)
564 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
566 changes in sbcl-1.0.1 relative to sbcl-1.0:
567 * new platform: FreeBSD/x86-64, including support for threading.
568 * new feature: the compiler stores cross-referencing information
569 about function calls (who-calls), macroexpansion (who-macroexpands)
570 and special variables (who-binds, who-sets, who-references) for code
571 compiled with (< SPACE 3). This information is available through the
572 sb-introspect contrib.
573 * new feature: users may subclass SEQUENCE, and have instances of
574 these classes interoperate with standard Common Lisp functions if
575 a number of methods are defined. (This feature is experimental
576 and the interface subject to change based on feedback from SBCL
577 users and the general community)
578 * improvement: sb-sprof traces call stacks to an arbitrary depth on
579 x86/x86-64, rather than the previous fixed depth of 8
580 * improvement: another pthread back-end of mutex "pthread-futex".
581 * bug fix: non-ascii command-line arguments are processed correctly
582 (thanks to Yaroslav Kavenchuk)
583 * bug fix: non-required arguments were not passed correctly when a method
584 defined using DEFMETHOD was called from a mop-generated method using
585 CALL-NEXT-METHOD (reported by Pascal Costanza)
586 * bug fix: recursion is now permitted in accessors through
587 SLOT-UNBOUND. (reported by Pascal Costanza)
588 * bug fix: an error was signaled at startup if the HOME environment
589 variable was defined, but had an empty value (reported by Peter Van Eynde)
590 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
591 signal a PROGRAM-ERROR, not a BUG.
592 * bug fix: SB-POSIX:READDIR works when built with large file support.
593 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
595 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
596 for better automatic indentation support. (thanks to Matt Pillsbury)
597 * optimization: loading generic functions no longer takes O(n^2) time,
598 proportional to the amount of methods in the generic function
599 (reported by Todd Sabin and Jeremy Brown)
600 * optimization: the FIND and POSITION family of sequence functions
601 are significantly faster on arrays whose element types have been
603 * improvements to the Windows port:
604 ** Intermittent heap corruption problems have been fixed. (thanks
605 to Alastair Bridgewater)
606 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
608 ** Lisp is able to unwind foreign exception frames from alien
609 callbacks. (thanks to Alastair Bridgewater)
611 changes in sbcl-1.0 relative to sbcl-0.9.18:
612 * improvement: experimental support for threading on FreeBSD/x86.
613 (thanks to NIIMI Satoshi)
614 * improvement: runtime option --dynamic-space-size can be used
615 to set the size of the dynamic space reserved on startup.
616 * improvement: floating point modes in effect are now saved in
617 core, and restored on startup.
618 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
619 startup, not time since first call to GET-INTERNAL-REAL-TIME.
620 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
621 threads are not running after *SAVE-HOOKS* have run.
622 * improvement: writes to CLOS instance slots are type-checked in code
623 compiled with (SAFETY 3)
624 * improvement: floating-point exception handling on FreeBSD (thanks to
626 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
627 (thanks to Zach Beane)
628 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
630 * improvement: added support for the Shift-JIS external format.
631 (contributed by NIIMI Satoshi)
632 * improvement: callbacks are supported on Linux/PPC. (thanks to
634 * bug fix: compiler bug triggered by a (non-standard) VALUES
635 declaration in a LET* was fixed. (reported by Kaersten Poeck)
636 * bug fix: file compiler no longer confuses validated and already
637 dumped structurres. (reported by Kaersten Poeck)
638 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
639 fill-pointer signals a type-error as required. (thanks to
641 * bug fix: disassemly of funcallable instances works.
642 * bug fix: single stepping on PPC.
643 * bug fix: fix thread-safety problems in the type system (generally
644 manifesting as nonsensical errors like "STRING is a bad type specifier
645 for sequences" or "The value 1 is not of type FIXNUM")
646 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
647 (reported by Josip Gracin)
648 * bug fix: an error is signaled for attempts to displace arrays with
649 incompatible element types (thanks to Mario Mommer)
650 * bug fix: more correct handling of wide characters in debug info
651 (bug reported by Attila Lendvai and fixed by Juho Snellman)
652 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
653 and don't cause extra consing
654 * optimization: MAP and MAP-INTO are significantly faster on vectors
655 whose elements types have been declared.
656 * Improvements to SB-SPROF:
657 ** Support for allocation profiling
658 ** Reduced profiling overhead, especially for long profiling runs
659 * Improvements to the Windows port:
660 ** floating point exceptions are now reported correctly.
661 ** stack exhaustion detection works partially.
662 ** more accurate GET-INTERNAL-REAL-TIME.
663 ** better RUN-PROGRAM behaviour in case of errors in the child
665 ** PROBE-FILE now simplifies pathnames correctly.
666 ** DIRECTORY now works correctly with :WILD-INFERIORS.
667 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
668 1st 1970 (although time timezone data will be unreliable).
669 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
671 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
672 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
673 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
675 * bug fix: two potential GC deadlocks affecting threaded builds.
676 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
677 returns T (reported by Anton Kazennikov)
678 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
679 with non-variable places
680 * bug fix: remove a race condition in the setting of
681 funcallable-instance functions, this should make threaded CLOS
682 code more stable against memory faults.
683 * bug fix: corruption of specials when unbinding is interrupted by an
684 asynchronous unwind (reported by Hannu Koivisto)
685 * improvement: the debugger will now also display local variables that
686 are only used once, for code compiled with a DEBUG optimization quality
689 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
690 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
691 * incompatible change: External-format support for FFI calls. The
692 SB-ALIEN:C-STRING no longer implies an ASCII
693 external-format. Instead the string is subject to external-format
694 conversion. Additionally return values of type SB-ALIEN:C-STRING
695 are by default of type (SIMPLE-ARRAY CHARACTER), not
696 SIMPLE-BASE-STRING. For an alien type that behaves like the old
697 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
698 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
699 * incompatible change: SB-EXT package no longer contains the
700 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
701 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
703 * incompatible change: the single-stepper is no longer available
704 on Alpha, Mips and Sparc platforms.
705 * minor incompatible change: the direct superclasses of
706 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
707 not (STANDARD-OBJECT FUNCTION). This makes the
708 class-precedence-lists of GENERIC-FUNCTION and
709 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
711 * minor incompatible change: the default stream external format on
712 non-unicode SBCL is no longer detected from the locale, but is always
714 * new feature: Add a version of evaluator that uses an interpreter instead
715 of the compiler. EVAL still uses the compiler by default, to switch it
716 to use the interpreter, set the value of the variable
717 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
718 * minor incompatible change: the single-stepper REPL has been merged
719 with the normal debugger (see the "Stepping" heading of the debugger help
720 for more details). The debugger command STEP will no longer switch
721 to the single-stepper REPL.
722 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
723 for a type now works.
724 * bug fix: loading of default sysinit file works. (thanks to Leonid
726 * bug fix: better detection of circularities in the file-compiler.
727 (reported by Marco Monteiro)
728 * bug fix: the CL pathname functions now work with files that have
729 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
730 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
731 systems with Unix98 pty semantics.
732 * bug fix: ASDF-INSTALL will now work with bsd tar.
733 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
735 * bug fix: timers expiring in dead threads no longer cause a
736 type-error (reported by Paul "Nonny Mouse").
737 * bug fix: thanks to more lightweight single-stepper instrumentation,
738 code compiled with (DEBUG 3) will compile and execute significantly faster,
739 and will have more accurate type-inferencing than before
740 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
741 whose bindings are modified
742 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
743 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
744 ** replace dummy implementations of CL:MACHINE-INSTANCE and
745 CL:SOFTWARE-VERSION with proper version
747 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
748 * feature: implemented the READER-METHOD-CLASS and
749 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
750 as specified by AMOP.
751 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
753 * optimization: faster LOGCOUNT implementation on x86 and x86-64
754 (thanks to Lutz Euler)
755 * optimization: hashing of general arrays and vectors has been
756 improved. (reported by Any Fingerhut)
757 * enhancement: SB-INTROSPECT is now able to find definitions of
758 profiled functions. (thanks to Troels Henriksen)
759 * enhancement: compiler-macro expansion applies now to FUNCALL forms
760 as well, allowing compiler-macros for SETF-functions to expand.
761 * enhancement: step-instrumentation no longer wraps known
762 single-value functions with multiple-value context, allowing
763 better type inference.
764 * fixed bug #337: use of MAKE-METHOD in method combination now works
765 even in the presence of user-defined method classes. (reported by
766 Bruno Haible and Pascal Costanza)
767 * fixed bug #339(c): if there are applicable methods not part of any
768 long-form method-combination group, call INVALID-METHOD-ERROR.
769 (reported by Bruno Haible)
770 * fixed bug #361: the :FUNCTION initarg in the protocol for
771 initialization of methods can now be used to override
772 internally-produced optimized functions. (reported by Bruno
774 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
775 system-provided lambda expression no longer cause warnings about
776 unbound #:|pv-table| symbols.
777 * bug fix: improved the handling of type declarations and the
778 detection of violations for keyword arguments with non-constant
780 * bug fix: potentially erronous calls to PATHNAME and
781 MERGE-PATHNAMES were being flushed in some cases.
782 (reported by Richard Kreuter)
783 * bug fix: compiled calls to TYPEP were mishandling obsolete
784 instances. (reported by James Bielman and Attila Lendvai)
785 * bug fix: format strings with ~> without matching ~< no longer
786 trigger an AVER, but signal an understandable error instead.
787 (reported by Antonio Martinez)
788 * bug fix: specifying an output-file without a pathname-type for
789 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
790 of input-file instead of "fasl". (reported by Robert Dodier)
791 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
792 (reported by James Y Knight).
793 * bug fix: compiler-macros-function did not consider the environment
794 argument for shadowing by local functions.
795 * bug fix: compiler-macros expansion was inhibited by local INLINE
797 * bug fix: inline expansions of known functions were subject to
798 step-instrumentation in high debug policies, leading to problems
800 * bug fix: compiler failed to differentiate between different CONS
802 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
803 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
804 * bug fix: #368: incorrect use of expressed vs. upgraded array
806 * bug fix: #306a: more precise unions of array types.
807 * thread-safety improvements:
808 ** CONDITION-WAIT could return early on Linux, if the thread was
809 interrupted and subsequently continued with SIGCONT.
810 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
812 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
813 * added support for the ucs-2 external format. (contributed by Ivan
815 * minor incompatible change: pretty printing of objects of type
816 (cons symbol) is, in the default pprint-dispatch-table, now
817 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
819 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
820 called later in a class's lifetime, possibly as late as when the
821 first instance of the class is created. Previously,
822 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
823 class became finalizeable.
824 * fixed bug: FILE-POSITION sometimes returned inconsistent results
825 for multibyte external-format streams. (thanks to "vbzoli")
826 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
827 with :ALLOCATION :CLASS inherited from superclasses of the
829 * fixed bug: anonymous classes can now be created using the :NAME
830 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
831 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
832 * fixed bug: core-files saved with :EXECUTABLE T can again be
833 executed when SBCL_HOME isn't set. (reported by James Knight)
834 * fixed bug: toplevel LOCALLY forms with declarations could
835 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
836 * fixed bug: printing from several different threads using different
837 values of *print-case* could cause invalid output, due to
838 some internal special variables of the printer not being bound
839 thread-locally (reported by Max Mikhanosha)
840 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
841 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
842 (reported by Pascal Costanza)
843 * fixed bug: FINALIZE-INHERITANCE is called from
844 REINITIALIZE-INSTANCE on classes when the class has previously
845 been finalized, as required by AMOP.
846 * minor code generation optimizations:
847 ** better register allocation in CLOS dispatching functions
848 ** overflow detection when coercing signed bytes to fixnums on x86-64
849 ** is now implemented with one IMUL instruction instead of three shifts
850 ** more efficient bit-vector access on x86 and x86-64
851 ** more efficient access to raw structure slots on x86 and x86-64
852 * fixed some bugs revealed by Paul Dietz' test suite:
853 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
856 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
857 * feature: thread support on Solaris/x86, and experimental thread support
859 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
861 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
862 default method for SLOT-UNBOUND.
863 * minor incompatible change: prevent the user from specializing the
864 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
865 somewhat counter to the intent of the protocol, I (CSR) think, and
866 additionally it just doesn't work in SBCL as currently
867 implemented, thanks to optimizations (that are always valid for
868 the other three SLOT-VALUEish functions, but not for the setter).
869 * minor incompatible change: the :SB-LDB feature is now enabled by
870 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
871 the low-level debugger.
872 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
873 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
874 down to an SB-SYS:FD-STREAM, allowing direct communication in
875 more cases. Temporary files and pipes are still used as a fallback
877 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
878 on the non-reentrant C functions localtime(3) and gmtime(3).
879 * bug fix: global optimization policy was not visible in LOCALLY and
881 * bug fix: class objects can be used as specializers in methods.
882 (reported by Pascal Costanza)
883 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
884 with a type error. (reported by blitz_ on #lisp)
885 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
886 failed with a type error. (reported by Pascal Bourguignon)
887 * bug fix: merging pathnames against defaults with :DIRECTORY
888 starting with '(:RELATIVE :BACK) should preserve the :BACK.
889 (reported by James Y Knight)
890 * bug fix: saving large (>2GB) cores on x86-64 now works
891 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
892 constant index and a (simple-array (signed-byte 32)) array
893 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
894 error (patch by Robert J. Macomber)
895 * feature: TIME now displays the amount of run-time spent in GC
896 * bug fix: The debugger now does a better job of respecting
897 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
898 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
899 workaround for bug 403.)
900 * bug fix: floating point exception handling now works on FreeBSD
901 (thanks to NIIMI Satoshi)
902 * fixed some bugs revealed by Paul Dietz' test suite:
903 ** MISC.641: LET-conversion were not supposed to work in late
905 * improvements to the Win32/x86 port:
906 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
907 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
909 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
910 * new feature: source path information is generated for macro-expansion
911 errors for use in IDE's like Slime (thanks to Helmut Eller)
912 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
913 * bug fix: compiler does not loop forever on an invalid type in
915 * improvement: compilation of most CLOS applications is significantly
917 * optimization: added a limited bytecode compiler for simple toplevel
918 forms, speeding up compilation and FASL loading
919 * bug fix: the statistical profiler now properly distinguishes anonymous
922 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
923 * minor incompatible change: in sbcl-0.9.11 (but not earlier
924 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
925 into subclasses of the finalized class. Now user calls to
926 FINALIZE-INHERITANCE finalize just the one class, and calls by the
927 system return before any subclasses are finalized.
928 * minor incompatible change: The reader no longer ignores errors
929 regarding non-existent packages in #+ and #- feature tests.
930 * new feature: command line options --no-sysinit, --no-userinit to
931 inhibit loading the corresponding init files
932 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
933 for sending data through UDP sockets (thanks to François-René Rideau)
934 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
935 error is signalled instead (thanks to François-René Rideau)
936 * minor incompatible change: Error signalling behaviour of lexical
937 operations violating package locks has changed slightly. Refer to
938 documentation on package locks for details.
939 * enhancement: EVAL can process IF-expressions without resorting to the
941 * bug fix: Some operations involving SIN, COS, or TAN weren't being
942 constant-folded properly due to missing out-of-line functions.
943 (reported by Mika Pihlajamäki)
944 * bug fix: LISTEN sometimes returned T even in cases where no data was
945 immediately available from the stream
946 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
947 were reversed. (reported by Levente Mészáros)
948 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
949 when given a vector displaced to an adjustable array. (reported
951 * bug fix: derivation of float boundaries from numbers outside the
952 appropriate float range (reported by John Wiseman)
953 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
955 * bug fix: precomputing cache entries for generic functions with
956 some subclasses of specializers as yet invalid does not attempt to
957 fill a cache line with a negative offset. (reported by Levente
959 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
960 allow condition class objects as arguments to DOCUMENTATION and
961 (SETF DOCUMENTATION); only find and set documentation for
962 structure names for the STRUCTURE doc-type. (suggested by Gary
964 * improvements to the Win32/x86 port:
965 ** (user-homedir-pathname) and default initialization file
966 locations now know about the user's "Documents and Settings"
967 directory (thanks to Yaroslav Kavenchuk)
968 ** run-program is implemented (thanks to Mike Thomas)
969 ** sockets support (thanks to Timothy Ritchey)
970 ** better backtrace support (thanks to Alastair Bridgewater)
971 ** sb-grovel supported
972 ** asdf-install and sb-posix work somewhat
973 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
974 * improvements to the Solaris/x86 port:
975 ** works on Solaris 11/Solaris Express
976 ** floating-point exception handling support
977 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
978 * fixed some bugs revealed by Paul Dietz' test suite:
979 ** REMOVE-METHOD returns its generic function argument even when
980 the method is not one of the generic functions' methods.
981 ** objects declared as MEMBER types can be admissible arguments to
983 ** printing characters should simply be printed by the FORMAT ~:C
985 ** compiler failure when compiling functions with hairy constant
986 defaults for optional parameters.
987 ** compiler produces wrong code when MAYBE-INLINE-expanding a
988 function, which is already optimized.
990 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
991 * new platform: experimental support for SBCL x86/Darwin, including
992 MacOS X 10.4.5 on Intel.
993 * new feature: Unicode character names are now known to the system
994 (through CHAR-NAME and NAME-CHAR).
995 * new feature: the filesystem location of SBCL's core file is
996 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
997 * minor incompatible change: the contrib modules SB-POSIX and
998 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
999 this change is to make it easier to distribute
1000 location-independent binaries.
1001 * bug fix: as implied by AMOP, standardized classes no longer have
1002 slots named by external symbols of public packages. (reported by
1004 * optimization: calling functions via constant symbols -- as in
1005 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1006 via the function object as in (FUNCALL #'FOO).
1007 * optimization: on x86 and x86-64, the calling convention for the
1008 case of dealing with an unknown number of values has been altered
1009 to be friendlier to the prediction heuristics implemented,
1010 particularly with reference to CALL and RET pairing. (thanks to
1011 Alastair Bridgewater)
1012 * optimization: on x86, the code for access to thread-local symbol
1013 values and binding/unbinding of thread-local symbols is smaller.
1014 (thanks to Alastair Bridgewater)
1015 * enhancement: CONSTANTP is now able to determine constantness of
1016 more complex forms, including calls to constant-foldable standardized
1017 functions and some special forms beyond QUOTE.
1018 * fixed bug: occasional GC crashes on Solaris/x86
1019 * optimization: x86-64 supports stack allocation of results of simple
1020 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1021 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1022 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1023 (thanks to James Knight)
1024 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1025 by the standard (thanks to Stephen Compall)
1027 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1028 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1029 be used for bundling the runtime and the core file into one
1030 executable binary. This feature is not currently supported on all SBCL
1031 platforms. (thanks to James Bielman and NIIMI Satoshi)
1032 * new feature: a generational or ephemeral garbage collector is now
1033 the default on the PowerPC platform (both Linux and Darwin). The
1034 old Cheney (stop and copy) collector is a build-time option.
1035 (thanks to Cyrus Harmon, after Raymond Toy)
1036 * minor incompatible change: the method by which SBCL finds its
1037 contributed modules has changed; it no longer relies on symbolic
1038 links from an $SBCL_HOME/systems directory, but searches directly
1039 in the subdirectories of $SBCL_HOME.
1040 * enhancement: the dynamic heap size on the Linux/PPC platform is
1041 markedly larger, even using the older Cheney garbage collector.
1042 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1043 certain complicated string types. (reported by Gary King)
1044 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1045 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1046 * fixed bug: DOCUMENTATION on structures no longer signals an
1047 error if no structure type of the right name exists. (reported by
1049 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1050 applicable non-standard primary or :AROUND method on
1051 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1052 no longer cause unbound variable errors. (reported by Kevin Reid)
1053 * fixed bug: metacircle resolution in cases where methods have slots
1054 added before the slots from STANDARD-METHOD. (reported by Jean
1056 * fixed bug: the Power PC assembly code for calling into lisp
1057 sometimes computed its return address wrongly (depending on the
1058 15th bit of the address, fixed at link time). This used to
1059 manifest itself in a segmentation violation while building PCL.
1060 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1061 many others over the years)
1062 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1063 the stack pointer (treated as a signed integer) is negative.
1064 (thanks to Peter van Eynde)
1066 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1067 * new platform: experimental support for the Windows operating
1068 system has been added. (thanks to Alastair Bridgewater)
1069 * fixed several bugs in and robustified the PPC FFI (including
1070 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1071 * bug fix: allow non-simple string symbol names (reported by Paul
1073 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1075 * bug fix: saving a core corrupted callbacks on x86/x86-64
1076 * bug fix: closed a loophole in metacircularity detection and
1077 grounding in the PCL implementation of CLOS.
1078 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1079 (reported by tomppa on #lisp)
1080 * bug fix: clear the direction flag on Lisp -> C transitions, as
1081 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1082 (reported by Andrej Grozin and Hendrik Maryns)
1083 * optimization: major improvements to GC efficiency on GENCGC platforms
1084 * optimization: faster implementation of EQUAL
1085 * optimization: emit more efficient opcodes for some common
1086 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1087 * optimization: several other minor code-generation improvements on x86-64
1088 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1090 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1091 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1092 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1093 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1094 by ANSI as a generic function, and for consistency (SETF
1095 GENERIC-FUNCTION-NAME) is treated likewise.
1096 * fixed bug #233.b: make constraint propagation notice when a variable
1097 value is changed after it is referenced but before it is used
1098 * fixed bug #296: no more arbitrary behaviour from filenames with
1099 odd characters as --load arguments. (reported by Adam Warner)
1100 * fixed bug #390: :CHARACTER-SET pathname components now work as
1101 expected. (reported by Tim Daly Jr)
1102 * fixed bug #391: complicated :TYPE intersections in slot
1103 definitions no longer cause an error in PCL internals.
1104 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1105 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1106 returning the number of octets which would be written to the
1107 file-stream. (thanks to Robert J. Macomber)
1108 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1109 arguments to the internal stream functions. (thanks to David
1111 * bug fix: stack exhaustion is now handled on the x86/SunOS
1112 platform. (thanks to Daisuke Homma)
1113 * bug fix: more accurate ROOM results on GENCGC platforms
1114 * optimization: improved type inference for arithmetic-for
1115 index variables in LOOP
1116 * optimization: faster floating-point SQRT on x86-64
1117 * fixed some bugs revealed by Paul Dietz' test suite:
1118 ** DOCUMENTATION returns NIL instead of "" for method combinations
1119 that don't have a docstring
1121 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1122 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1123 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1124 therefore are not customizeable by user code (as seems to be at
1125 least permitted and maybe required by AMOP). As a consolation,
1126 however, the SBCL implementation of these functions now calls
1127 REINITIALIZE-INSTANCE as specified by AMOP.
1128 * bug fix: slot-definition documentation is propagated as per ANSI
1129 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1130 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1131 argument, as do eslotd initialization methods. (from Pascal
1132 Costanza's "Closer" project)
1133 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1134 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1136 * bug fix: it is now possible to have more than one subclass of
1137 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1138 (reported by Bruno Haible, Pascal Costanza and others)
1139 * bug fix: the dependent update protocol now works for generic
1140 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1141 and Pascal Costanza)
1142 * bug fix: condition-class instances corresponding to
1143 DEFINE-CONDITION forms are now created eagerly. (reported by
1144 Kalle Olavi Niemitalo on comp.lang.lisp)
1145 * bug fix: floating point printing is more accurate in some
1146 circumstances. (thanks to Simon Alexander)
1147 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1148 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1149 * bug fix: callbacks on OS X now preserve stack-alignment.
1150 (thanks to Cyrus Harmon)
1151 * enhancement: the x86-64 disassembler is much better at
1152 disassembling SSE instructions. (thanks to Lutz Euler)
1153 * enhancement: CHECK-TYPE on a variable now allows type-inference
1154 (thanks to Vincent Arkesteijn)
1155 * optimization: improved performance of EUC-JP external format.
1156 (thanks to NIIMI Satoshi)
1157 * optimization: performance improvements to IO on file streams of
1158 :ELEMENT-TYPE CHARACTER
1159 * optimization: much faster memory allocation on x86-64
1160 * optimization: faster garbage collections (latency and throughput)
1162 * optimization: faster bignum subtraction and division on x86 and x86-64
1163 (thanks to Lutz Euler)
1164 * optimization: more accurate type-inference for loops that use a
1165 floating point index variable or a negative step.
1167 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1168 * bug fix: add a workaround to SBCL looping infinitely at startup on
1169 Linux kernels with apparently buggy implementations of personality().
1170 (thanks to Svein Ove Aas)
1171 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1172 (thanks to David Lichteblau)
1173 * bug fix: local bindings shadow global symbol macros for
1174 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1175 * optimization: non-open coded uses of numeric comparison operators
1176 (e.g. >) no longer cons when called with more than one parameter
1177 on platforms supporting dynamic-extent allocation.
1178 * enhancement: saving cores with foreign code loaded is now supported
1179 on MIPS/Linux in addition to the previously supported platforms.
1180 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1181 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1182 * bug fix: division by zero in sb-sprof when no samples were collected
1183 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1184 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1185 (thanks to Alexander Kjeldaas)
1186 * new feature: ignoring the return values of destructive functions
1187 like NREVERSE signals a compile-time style-warning.
1188 (thanks to Kevin Reid)
1189 * new feature: the alignment of alien structure fields can be
1190 explicitly specified. (thanks to Cyrus Harmon)
1191 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1192 and dump core on SIGQUIT
1194 ** incompatible change: threads do not inherit values of specials
1195 from their parents (see manual)
1196 ** bug fix: threads stacks belonging to dead threads are freed by the
1197 next exiting thread, no need to gc to collect thread stacks anymore
1198 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1199 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1200 ** bug fix: dynamic variable and thread start related gc lossage
1201 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1203 * fixed some bugs revealed by Paul Dietz' test suite:
1204 ** SUBTYPEP is slightly more accurate on heinously complicated
1205 CONS types where some of the members have uncertain (in the
1206 NIL, NIL sense) type relationships to each other.
1208 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1209 no longer purified unless :PURIFY T is explicitly specified.
1210 ** Non-purified cores are significantly smaller than before
1212 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1213 * new feature: timers based on Zach Beane's excellent timer package
1214 * added support for the following external formats: koi8-u,
1215 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1216 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1217 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1218 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1219 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1220 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1221 * incompatible change: a threaded SBCL will no longer revert to
1222 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1223 * bug fix: interrupts are disabled until startup is complete; no
1224 more sigsegvs when receiving a signal to soon
1225 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1227 * bug fix: add a workaround for the memory randomization features in
1228 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1229 workaround will only be in effect on systems with the proc filesystem
1231 * bug fix: printing objects of type HASH-TABLE signals a
1232 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1234 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1235 have been running for over 50 days. (reported by Gilbert Baumann)
1236 * bug fix: the logic for getting names of functions gets less
1237 confused when confronded with alternate-metaclass
1238 funcallable-instances. (reported by Cyrus Harmon)
1239 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1240 consistent, even on internal alternate-metaclass objects.
1241 * bug fix: flush closure information collected by physical
1242 environment analysis prepass before the main pass. (bug reported
1244 * bug fix: compiler pack phase does not modify a hashtable, which is
1245 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1246 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1247 may be expanded. (reported by Alexander <malishev> on c.l.l)
1248 * bug fix: no more style-warnings when compiling code using LAST
1249 under high-SPEED low-SPACE settings. (reported by David Wallin
1251 * fixed some incompatibilities between SBCL's MOP and the MOP
1253 ** the METAOBJECT class is now implemented;
1254 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1255 STANDARD-OBJECT, as required;
1256 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1257 now compatible (as required by VALIDATE-SUPERCLASS); there
1258 remains a consistency requirement over the presence or absence
1259 of the FUNCTION class in the superclass list of finalized
1260 classes; see the manual for more details;
1261 ** the :AROUND method for COMPUTE-SLOTS on
1262 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1263 requested slot ordering.
1265 ** bug fix: parent thread now can be gc'ed even with a live
1267 ** bug fix: binding a special with PROGV to no value is not
1268 the same as that symbol not having been bound (thanks to
1270 ** bug fix: binding specials is thread safe (thanks to
1272 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1273 and SB-SPROF do not die with 'no handler for signal XX in
1274 interrupt_handle_now(..)' anymore
1275 ** bug fix: WITH-TIMEOUT works with multiple running threads
1276 * fixed some bugs revealed by Paul Dietz' test suite:
1277 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1278 the :method-class keyword argument.
1280 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1281 * new port: the Solaris operating system on x86 processors is now
1282 mostly supported, though some rough edges in the environment
1283 remain. (thanks to Daisuke Homma)
1284 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1285 space for its heap. (thanks to Thiemo Seufer)
1286 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1287 closures. (thanks to Thiemo Seufer)
1288 * minor incompatible change: eof selects abort in the debugger.
1289 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1290 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1292 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1293 deprecated, and will go away in a future revision of SBCL.
1294 * minor incompatible change: GC-ON and GC-OFF are no longer
1295 implemented with a counter, it does not matter how many times gc
1296 is switched on or off
1297 * bug fix: discriminating functions for generic function classes
1298 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1299 make invalid assumptions about method precedence order. (reported
1301 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1302 platforms. (thanks to Thiemo Seufer)
1303 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1304 * bug fix: correct pathname printing with printer escaping is on.
1305 (thanks to Kevin Reid)
1306 * bug fix: complex VOP definitions in "user-space" no longer trigger
1307 package locks. (reported by Zach Beane)
1308 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1309 for funcallable-instance functions, and is no different from
1311 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1312 characters in the current readtable. (reported by Nicholas Neuss)
1313 * bug fix: SUBTYPEP on various CONS types returns more a more
1314 accurate acknowledgment of its certainty.
1315 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1316 lists in some cases. This partially fixes bug 384.
1317 * flush all standard streams before prompting in the REPL and the
1319 * bug fix: signal handling and triggering gc do not conflict
1320 directly anymore, in particular a high frequency sb-sprof does
1321 not prevent gc from running
1322 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1323 approximation for timezone and DST information between the
1324 universal time epoch and the smallest negative 32-bit time_t.
1325 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1326 year 1899 so long as the encoded time is non-negative.
1327 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1328 the correct answer for non-constant rotate amounts when performing
1329 an inline 32-bit rotation.
1331 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1332 there is only one thread in the session
1333 ** bug fix: memory leak for streams created in one thread and
1334 written to in another
1335 ** bug fix: lockup when compiled with gcc4
1336 ** bug fix: race that allows the gc to be triggered when gc is
1338 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1339 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1341 ** bug fix: finished threads are gc'ed properly
1342 * fixed some bugs revealed by Paul Dietz' test suite:
1343 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1344 has arguments with a different set of applicable methods from
1345 the orignal arguments.
1346 ** The type error thrown by MAP now has a correct expected-type
1348 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1349 name a compiled function.
1350 ** Three MISC tests where a large bignum was improperly coerced to
1351 a float that couldn't represent that bignum during type
1352 derivation were fixed.
1353 ** SUBTYPEP can now handle types involving the negation of a
1354 list-form FUNCTION type.
1355 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1356 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1357 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1359 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1360 * New feature: Experimental support for bivalent streams: streams
1361 opened with :element-type :default now allow character and binary
1362 (unsigned-byte 8) I/O
1363 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1364 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1366 * bug fix: no more highly sporadic "couldn't check whether ~S is
1367 readable" when reading a stream and an interrupt hits in the middle
1368 of a select system call
1369 * compiler better recognizes complex arrays (reported by Eduardo
1371 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1372 for variables on SBCL built with linkage-tables. (reported by Luis
1374 * various error reporting improvements.
1375 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1376 (Thanks to Thiemo Seufer)
1377 * bug fix: nested reader invokations work correctly
1378 * bug fix: it is possible to have simultaneous references to foreign
1379 code and foreign data with the same name.
1381 ** added x86-64 support
1382 ** incompatible change: the threading api now works with thread
1383 objects instead of thread ids
1384 ** bug fix: threads are protected from signals and interruption when
1385 starting up or going down
1386 ** bug fix: a race where an exiting thread could lose its stack to gc
1387 ** bug fix: don't halt on infinite error in threads if possible
1388 ** fixed numerous gc deadlocks introduced in the pthread merge
1389 ** bug fix: fixed thread safety issues in read and print
1390 ** bug fix: debugger doesn't hang on session lock if interrupted at
1391 an inappropriate moment
1392 ** bug fix: run-program is now thread safe(r)
1393 ** bug fix: inner with-recursive-lock no longer releases the mutex
1394 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1395 * fixed some bugs revealed by Paul Dietz' test suite:
1396 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1397 EXPECTED-TYPE slots filled.
1398 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1400 ** MAKE-CONDITION accepts classes as type-designators.
1401 ** COMPILE may never return NIL.
1402 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1403 range before calling Unix time functions
1405 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1406 * numerous signal handling fixes to increase stability
1407 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1408 * minor incompatible change: we now correctly canonize default
1409 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1410 as per the MOP, rather than the historical (INITARG INITFUNCTION
1411 INITFORM). (reported by Bruno Haible)
1412 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1413 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1414 for more information.
1415 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1416 pathname is a directory pathname.
1417 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1418 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1420 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1421 :METACLASS argument in addition to a class name. (reported by
1422 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1423 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1424 (reported by Bruno Haible; patch from Gerd Moellmann)
1425 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1427 * bug fix: more cleanups to the floating point exception handling on
1428 x86-64 (thanks to James Knight)
1429 * bug fix: the compiler does not try to constant fold calls of
1430 COERCE and BIT-* functions when they return freshly constructed
1431 objects. (reported by Nikodemus Siivola and Paul Dietz)
1432 * optimization: improved the MIPS versions of generic (in the
1433 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1434 * optimization: direct conversion of (unsigned-byte 32) to floats on
1435 the PowerPC platform.
1436 * optimization: structure instances with raw slots now use less
1437 memory, and probably show better memory locality. (thanks to
1439 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1440 are treated as requests for stack allocation on the x86-64,
1441 Alpha-32, PPC and SPARC platforms.
1442 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1443 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1444 (thanks to Tony Martinez)
1446 ** gcing a dead thread can no longer lead to lockups
1447 ** threads block signals until they are set up properly
1448 ** errno is no longer shared by threads
1449 ** interrupt-thread restores the eflags register on x86
1450 ** fixed some lockups due to gc/thread interaction
1451 * fixed some bugs revealed by Paul Dietz' test suite:
1452 ** invalid dotted lists no longer raise a read error when
1453 *READ-SUPPRESS* is T
1454 ** signal an error if a symbol that names a declaration is used
1455 as the name of a type, or vice versa
1456 ** allow using the (declare (typespec var*)) abbreviation for
1457 (declare (type typespec var*)) with all type specifiers
1458 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1459 ** always return NIL from PPRINT-POP when OBJECT is NIL
1460 ** don't signal errors when pretty-printing malformed LABELS,
1461 FLET or MACROLET forms
1462 ** declarations in a DOLIST body are also in scope for the
1464 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1466 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1469 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1470 * fixed cross-compiler leakages that prevented building a 32-bit
1471 target with a 64-bit host compiler.
1472 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1473 opened with :IF-EXISTS :OVERWRITE.
1474 * fixed bug 281: error for an invalid qualifier in a short-form method
1475 combination method is not signalled until the faulty method is called.
1476 * bug fix: iteration variable type inferrer failed to deal with open
1477 intervals. (reported by Alan Shields)
1478 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1479 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1480 or more runtime options were provided to the sbcl binary.
1481 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1482 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1483 * optimization: there's now a fast-path for fixnum arguments in the
1484 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1485 * optimization: the code generated on x86-64 is more compact thanks
1486 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1487 * bug fix: floating-point exception handling now partly works on
1488 x86-64. (Thanks to James Knight)
1489 * improvement to the MIPS backend from Thiemo Seufer: C-style
1490 64-bit long long arguments and return values to alien functions
1492 * fixed some bugs revealed by Paul Dietz' test suite:
1493 ** the type-error signalled from WARN has a filled-in DATUM slot.
1494 ** the type-error required when a stream is not associated with
1495 a file has the stream as its datum.
1496 ** type-errors on single-floats on x86-64 no longer have
1497 :INVALID-OBJECT as the datum
1498 ** the type-errors signalled for invalid function names now have
1499 a correct expected type
1500 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1501 for structures defined with a :TYPE. Documentation strings for
1502 typed structures are no longer immediately discarded
1503 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1504 broadcast streams and synonym streams. FILE-LENGTH now also works
1505 on broadcast streams.
1507 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1508 * incompatible change: the --noprogrammer option, deprecated since
1509 version 0.7.5, has been removed. Please use the equivalent
1510 --disable-debugger option instead.
1511 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1513 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1514 inoperational for a while now) has been completely removed.
1515 * null lexical environments are now printed as #<NULL-LEXENV>,
1516 significantly reducing the amount of clutter in typical backtraces.
1517 * documentation on weak pointers, finalization, and after GC hooks
1518 has been added to the manual.
1519 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1520 as other specialized array types, is much faster. SUBSEQ and
1521 COPY-SEQ on such arrays have also been sped up.
1522 * optimization: EQL is now more efficient when at least other argument
1523 is known to be of type (OR FIXNUM (NOT NUMBER)).
1524 * fixed bug: the runtime is now less vulnerable to changes in the
1525 size of the SBCL object on OS X, and virtual memory is reserved for
1526 all spaces that need to be at a fixed address.
1527 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1528 * fixed bug: finalizers and after GC hooks that cause consing are now
1530 * fixed bug: compiler error messages and summaries are now printed to
1531 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1532 * fixed inference of the upper bound of an iteration variable.
1533 (reported by Rajat Datta).
1534 * fixed bug 211e: calling local functions with duplicated constant
1535 keyword argument no longer causes a bogus style warning about an
1537 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1538 ftype declaration to be disregarded. (reported by Dave Roberts)
1539 * fixed bug 373: caused by erronous compilation of references to alien
1540 variables in the runtime on ppc/darwin.
1541 * fixed bug 376: CONJUGATE type deriver.
1542 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1543 (reported by Baughn on #lisp)
1544 * a cross-compiler bug on non-x86 platforms has been identified and
1545 fixed. (thanks to Bruno Haible)
1546 * improvements to the MIPS runtime code for increased stability.
1547 (thanks to Thiemo Seufer)
1548 * increased the maximimum compact environment size to allow
1549 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1550 * partial workaround for bug 135: don't name the function we're
1551 calling for hairy cases of EVAL, so as not to accumulate one environment
1552 entry per call to EVAL. (reported by Kevin Reid)
1553 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1554 * optimization: type testing for non-vector arrays should be faster.
1555 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1557 * bug fix: setting 31st element of a bit vector to zero did not work
1559 * bug fix: redefining a class definition which failed due to a
1560 previous accessor / function clash now works (but see BUGS entry
1561 #380 for more problems in this area). (thanks to Zach Beane)
1562 * the long form of DEFINE-METHOD-COMBINATION disables method group
1563 checking when given a single method group with pattern *. (thanks
1565 * on x86 compiler supports stack allocation of results of simple
1566 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1567 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1568 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1569 * fixed some bugs related to Unicode integration:
1570 ** the restarts for recovering from input and output encoding
1571 errors only appear when there is in fact such an error to
1573 * fixed some bugs revealed by Paul Dietz' test suite:
1574 ** MISC.549 and similar: late transformation of unsafe type
1575 assertions into derived types caused unexpected code
1577 ** SCALE-FLOAT type deriver is less wrong.
1578 ** type derivers for EXP, LOG and similar functions compute result
1579 types for complex arguments better.
1580 ** (MISC.563) CONJUGATE type deriver works for very restricted
1582 ** out-of-line type testers for character strings are available.
1583 ** EQUAL compiler transform understands specialness of objects
1585 ** accessing double-floats stored on the stack now works on x86-64.
1586 ** debugger internals could sometimes create invalid lispobjs,
1587 resulting in GC crashes.
1588 ** MISC.548: type check weakening can convert required type into
1590 ** initialization forms for bindings are not in scope of free special
1593 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1594 * incompatible change: thread support for non-NPTL systems has
1595 been removed - locking is buggy and unreliable. A threaded
1596 SBCL build will now warn at startup and refuse to create
1597 new threads, unless futex support is detected in the kernel
1598 * incompatible change: the top level REPL now has only an ABORT
1599 restart associated with it, not TOPLEVEL and ABORT as it used to.
1600 TOP and TOPLEVEL are now available as debugger commands for
1601 returning to the top level.
1602 * incompatible change: forms evaluated in the REPL now use the
1603 global optimization policy.
1604 * incompatible change: user- and system-initialization files are
1605 no longer processed with LOAD, but by READ and EVAL; hence the
1606 global optimization policy, startup package, readtable, etc,
1608 * The .fasl file format number has been incremented because of
1609 various incompatible changes.
1610 * internal entry point details and argument counts no longer appear
1611 in backtraces unless explicitly requested by setting
1612 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1613 * built-in and standard functions no longer have names like "top
1614 level local call to FOO".
1615 * fixed bug 32: functions defined in non-null lexical environments
1616 now have more legible printed representation
1617 * fixed bug 33: functions defined in non-null lexical environemnts
1618 are now more amenable to inspection by INSPECT.
1619 * workaround for bug 354: XEPs no longer appear in backtraces unless
1620 explicitly requested.
1621 * fixed bug: receiving the signal which results from attempting to
1622 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1623 SIGBUS on other *BSDs) on architectures where the C stack is also
1624 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1625 to debugger confusion or wild instability. (reported by Cheuksan
1627 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1628 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1629 (reported by Lutz Euler)
1630 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1631 compiled code. (reported by Paul Dietz)
1632 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1633 specialized parameter to a method to something that is not TYPEP
1634 the specializer is now possible.
1635 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1636 face of package deletion.
1637 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1638 pathname. (reported by David Tolpin on c.l.l)
1639 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1640 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1641 STRING-STREAM and FILE-STREAM, respectively.
1642 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1643 than previously; in addition, its test suite is now run on build.
1644 (thanks to Luke Gorrie)
1645 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1646 * optimization: added a immediate representation for single-floats
1648 * fixed some bugs revealed by Paul Dietz' test suite:
1649 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1650 ** attempting to create a package with a colliding nickname causes
1651 correctable errors to be signalled.
1652 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1653 ** several x86-64 backend bugs related to sign-extension of immediate
1656 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1657 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1658 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1660 * fixed loading of multiply forward-referenced layouts.
1661 (thanks to Cheuksan Wang)
1662 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1663 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1664 (see COMPILE-FILE documentation for details of :PRINT option)
1665 * fixed bugs 19 and 317: fixed-format floating point printing is
1666 more accurate. This also fixes a bug reported by Adam Warner
1667 related to the ~@F format directive.
1668 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1670 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1671 dispatch table if the from-char is a dispatch macro character.
1672 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1673 by the wordsize no longer ignore the last word. (reported by Lutz
1675 * fixed bug in type checking of dynamic-extent variables. (reported
1677 * optimization: sequence traversal functions use their freedom to
1678 coerce function designators to functions.
1679 * optimization: code with many calls to CLOS methods specialized on
1680 CLOS classes has had redundant type checks removed.
1681 * optimization: added declarations to speed up operations that access
1682 the internal character database (for example STRING-UPCASE)
1683 * optimization: comparison operations between floats and sufficiently small
1684 fixnums no longer create extra rationals
1685 * fixed some bugs related to Unicode integration:
1686 ** portions of multibyte characters at the end of buffers for
1687 character-based file input are correctly transferred to the
1688 start of the buffer at the next read.
1689 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1690 passing it through to OPEN.
1691 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1692 argument given to internal calls to OPEN.
1693 ** the built-in comment readers (introduced by character sequences
1694 ";" and "#|") are more forgiving to encoding errors; they will
1695 STYLE-WARN and then attempt to resync the stream at a character
1696 boundary. (thanks to Teemu Kalvas)
1697 * fixed some bugs in the x86-64 port:
1698 ** Negative short int return values from c-calls are sign-extended
1700 ** The stack is aligned to 16-bytes for c-calls, as required by
1701 the ABI. (thanks to Cheuksan Wang)
1702 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1703 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1704 for enums. (thanks to Vincent Arkesteijn)
1705 ** Multiple small optimizations and bugfixes for floating point
1707 * fixed some bugs revealed by Paul Dietz' test suite:
1708 ** Space, Tab, Linefeed, Return and Page have the invalid
1709 secondary constituent character trait.
1710 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1712 ** WITH-INPUT-FROM-STRING should only update the index place on
1714 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1715 works more reliably.
1716 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1717 with one-dimensional arrays with fill pointers.
1718 ** TYPE-OF failed on a complex with an integer realpart and a
1720 ** compiler failure during type inference for the code of form
1721 (IF (EQL X (THE ...)) ...) (MISC.535).
1723 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1724 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1725 platform. The port supports 61-bit fixnums, large memory spaces
1726 and reloading shared object files.
1727 * enhancement: saving cores with foreign code loaded is now
1728 supported on ppc/Darwin in addition to the previously supported
1730 * enhancement: the statistical profiler now walks deeper into the
1731 call stack for more meaningful call-graphs and accrued time
1732 reports (x86/x86-64 only). It also now reports time spent in
1734 * enhancement: it is now possible to trace most individual methods
1735 of a generic function in addition to tracing the generic function
1737 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1739 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1740 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1741 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1742 produces an error. (thanks to Vincent Arkesteijn)
1743 * bug fix: NAMESTRING on pathnames with :WILD components in their
1744 directories works correctly. (thanks to Artem V. Andreev)
1745 * fixed bug 125: compiler preserves identity of closures. (reported
1747 * bug fixed: functions with &REST arguments sometimes failed with
1748 "Undefined function" when compiled with (DEBUG 3). (reported by
1750 * bug fix: overflow during compiling of setting element of a bit
1751 vector with constant index and value. (reported by Timmy Douglas)
1752 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1753 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1754 * on x86 compiler partially supports stack allocation of dynamic-extent
1756 * GO and RETURN-FROM do not check the extent of their exit points
1757 when compiled with SAFETY 0.
1758 * fixed some bugs related to Unicode integration:
1759 ** encoding and decoding errors are now much more robustly
1760 handled; it should now be possible to recover even from invalid
1761 input or output to the terminal. (thanks to Teemu Kalvas)
1762 ** provided a first cut at implementing STRING-TO-OCTETS and
1763 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1764 ** altered the SB-MD5 contributed module slightly, changing the
1765 interface just enough for it to be supportable for builds where
1766 lisp characters are not eight bits.
1767 * fixed some bugs revealed by Paul Dietz' test suite:
1768 ** the FORMATTER-generated functions for ~V[ conditionals require
1769 the correct number of arguments.
1770 ** READ-FROM-STRING returns the mandated second value when applied
1771 to displaced strings.
1772 ** the #\Rubout and #\Backspace characters are treated as invalid
1773 constituent characters by the tokenizer.
1775 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1776 * new feature: reloading changed shared object files with
1777 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1778 * new feature: references to foreign variables and functions
1779 can now be compiled and loaded before the corresponding shared
1780 object file is loaded, as long as the foreign definitions are
1781 available at runtime.
1782 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1783 supported. (thanks to Dan Debertin)
1784 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1785 just keywords. (thanks to Vincent Arkesteijn)
1786 * fixed bug #331: structure-class instances corresponding to
1787 DEFSTRUCT forms are now created eagerly.
1788 * fixed bug #345: backtraces from calls to undefined functions work
1789 on x86 as well. Related bug #61 is now also partially fixed on x86
1790 (backtraces from throws to unknown catch tags.)
1791 * bug fix: lambda-list parsing is now stricter vrt. order and number
1792 of lambda-list keywords.
1793 * bug fix: as specified by AMOP, an error is signalled if a
1794 class-option appears multiple times in a DEFCLASS form. (reported
1796 * bug fix: RANDOM can be compiled when the compiler derives the type
1797 of its numeric argument as a disjoint set of small integers.
1798 (reported by Paul Dietz)
1799 * bug fix: starting a core saved with shared objects loaded when
1800 those objects are not available no longer causes threaded SBCL to
1801 hang. (reported by Sean Ross)
1802 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1804 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1805 argument, which must always be NIL. (reported by Kalle Niemitalo)
1806 * bug fix: printing 1.0d+23 no longer results in an error.
1807 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1808 * bug fix: structure slot setters preserve evaluation order. (thanks
1810 * fixed some bugs related to Unicode integration:
1811 ** RUN-PROGRAM can allow its child to take input from a Lisp
1812 stream. (reported by Stefan Scholl)
1813 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1814 locales. (reported by Ken Causey)
1815 * fixed some bugs revealed by Paul Dietz' test suite:
1816 ** INCF, DECF and REMF evaluate their place form as specified in
1818 ** FORMATTER expands ~{ ~} iteration directives with V or #
1819 parameters correctly.
1820 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1821 consequent uses no arguments correctly.
1822 ** the system has a partial understanding of the (COMPLEX RATIO)
1824 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1827 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1828 * new feature: a build-time option (controlled by the :SB-UNICODE
1829 keyword feature, enabled by default) for building the system with
1830 support for the entire 21-bit character space defined by the
1832 * new feature: the system now has rudimentary external-format
1833 support; the primary user-visible change at this time is that
1834 characters with the high bit set (such as lower-case-e-acute) will
1835 print correctly to a terminal in a UTF-8 environment.
1836 * minor incompatible change: BASE-CHAR no longer names a class;
1837 however, CHARACTER continues to do so, as required by ANSI.
1838 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1839 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1840 has been moved to the SB-EXT package (temporarily re-exported from
1842 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1843 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1845 * fixed bug #351: better error-handlind and reporting for malformed
1847 * fixed bug #350: bignum-printing is now more memory-efficient,
1848 allowing printing of very large bignums, eg. (expt 2 10000000).
1849 (reported by Bruno Haible)
1850 * fixed bug #302: better primitive-type selection for intersection
1852 * fixed bug #308: non-graphic characters now all have names, as
1853 required. (reported by Bruno Haible)
1854 * bug fix: redefining a class with different superclasses now correctly
1855 removes it from the direct-subclasses of its previous superclasses.
1856 (reported by David Morse)
1857 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1858 new value now works. (reported by Bruno Haible)
1859 * bug fix: correct canonicalization of multiple non-standard slot
1860 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1861 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1862 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1863 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1864 now exists, an signals an error.
1865 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1866 during macroexpansion for non-iteration variables as well. (reported
1867 by Bruno Haible for CMUCL)
1868 * bug fix: Cyclic structures and unprintable objects in compiler
1869 messages no longer cause errors. (reported by Bruno Haible)
1870 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1871 and READ-FROM-STRING all now return a primary value of NIL if
1872 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1873 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1874 (reported by Bruno Haible for CMUCL)
1875 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1876 do not designate a condition. (reported by Bruno Haible for
1878 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1879 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1880 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1881 permitting resolution in favour of any of the conflicting symbols.
1882 (reported by Bruno Haible for CMUCL)
1883 * bug fix: EQUAL compiler optimizations is less aggressive on
1884 strings which can potentially compare true despite having distinct
1885 specialized array element types.
1886 * bug fix: unit enumerations can be defined without dividing by
1887 zero. (thanks to Vincent Arkesteijn)
1888 * FORMAT compile-time argument count checking has been enhanced.
1889 (report from Bruno Haible for CMUCL)
1890 * a partial workaround for the bug 262: the compiler does not try to
1891 inline-expand a local function doing RETURN-FROM from a deleted
1892 BLOCK. (thanks to Peter Denno for the bug report and to David
1893 Wragg for the simple test case)
1894 * fixed some bugs revealed by Paul Dietz' test suite:
1895 ** RENAME-PACKAGE allows all package designators as new package
1897 ** constraint propagation and lambda variable substitution are
1898 more cautious in dealing with partially deleted code.
1899 ** compiler sometimes lost reoptimization passes.
1900 ** CERROR, given a condition as condition designator, passes its
1901 remaining arguments to the continue format control without
1903 ** Case-altering FORMAT directives work correctly on non-ASCII
1905 ** The REAL type specifier handles bounds outside the floating
1906 point ranges without signalling FLOATING-POINT-OVERFLOW.
1907 ** Functions with IR1-transformations can create intercomponent
1908 references to global functions.
1909 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1911 ** FORMAT ~R treats a nil value for its first parameter correctly.
1913 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1914 * enhancement: saving cores with foreign code loaded is now
1915 supported on x86/NetBSD and sparc/Linux in addition to the previously
1916 supported platforms.
1917 * bug fix: on some platforms repeated installations caused multiple
1918 copies of HTML documentation to be installed -- should not happen
1919 any more. (reported by Stefan Scholl)
1920 * bug fix: parsing self-recursive alien record types multiple times
1921 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1922 original patch by Helmut Eller for CMUCL)
1923 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1924 (thanks to Richard Kreuter)
1925 * bug fix: defining classes whose accessors are methods on existing
1926 generic functions in other (locked) packages no longer signals
1927 bogus package lock violations. (reported by François-René Rideau)
1928 * bug fix: special variables as DEFMETHOD parameters no longer have
1929 associated bogus type declarations. (reported by David Wragg and
1931 * bug fix: read-write consistency on streams of element-type
1932 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1933 * bug fix: redefiniton of the only method of a generic function with
1934 no DEFGENERIC no longer emits a full WARNING. In addition,
1935 redefinition of generic functions with no DEFGENERIC to an
1936 incompatible lambda list now signals an error. (thanks to Zach
1938 * bug fix: DEFGENERIC now works even when there's a function of the
1939 same name in an enclosing lexical environment. (thanks to Zach
1941 * fixed compiler failure, caused by instrumenting code during
1942 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1943 * optimization: added loop analysis and improved register allocation
1944 * fixed some bugs revealed by Paul Dietz' test suite:
1945 ** POSITION on displaced vectors with non-zero displacement
1946 returns the right answer.
1947 ** (SIMPLE-STRING) is a valid type specifier for sequence
1949 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1951 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1952 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1954 ** The justification version of the FORMAT ~< directive treats
1955 non-zero minpad parameter correctly.
1957 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1958 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1959 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1960 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1961 the supported interface.
1962 * new feature: Single-stepping of code compiled with DEBUG 2 or
1963 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1964 * new feature: saving cores with foreign code loaded is now
1965 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1966 Timothy Moore's work for CMUCL)
1967 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1968 parameters to * instead of NIL if no initform is supplied.
1969 (reported by Johan Bockgaard)
1970 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1971 correctly when FUNCALL appears as the car of the form. Note:
1972 despite this FUNCALL forms are not currently subject to
1973 compiler-macro expansion. (port of Raymond Toy's fix for the
1974 same from CMUCL, reported by Johan Bockgaard)
1975 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1976 (thanks for Teemu Kalvas)
1977 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1978 case instead of terminating the iteration (thanks for Julian
1979 Squires, Sean Champ and Raymond Toy)
1980 * bug fix: incorrect expansion of defgeneric that caused a style
1981 warning. (thanks for Zach Beane)
1982 * bug fix: slot accessor effective method computation works properly
1983 for classes with multiple non-standard applicable methods on
1984 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1985 * on x86 compiler supports stack allocation of results of LIST and
1986 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1987 CMUCL implementation by Gerd Moellmann)
1988 * fixed some bugs revealed by Paul Dietz' test suite:
1989 ** FORMAT strings with both the ~<~:;~> form of the justification
1990 directive and pretty-printing directives cause an error.
1992 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1993 * incompatible change: the internal functions
1994 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1995 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1996 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1997 instead of the old functions.
1998 * new feature: on platforms where "dladdr" is available foreign
1999 function names now appear in backtraces. (based on Helmut Eller's
2001 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2002 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2004 * documentation: ASDF manual is now distributed with SBCL.
2005 * bug fix: SBCL can now load its contributed modules with REQUIRE
2006 even if the system-provided entries have been removed from
2007 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2009 * bug fix: backtraces involving undefined functions or assembly
2010 routines are more informative. (thanks to Brian Downing)
2011 * bug fix: mutually referent alien structures now work correctly.
2012 (reported by Rick Taube)
2013 * bug fix: structures defined by WITH-ALIEN can be referred to
2014 within other definitions in the same WITH-ALIEN.
2015 * bug fix: division operators (MOD, TRUNCATE and the like) with
2016 constant zero divisors and integer dividends no longer generate
2018 * bug fix: provide default methods for INPUT-STREAM-P and
2019 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2020 * bug fix: improve the detection and resolution of MOP metacycles
2021 (where effective method computation depends on the generic function
2022 having its effective method computed). (reported by Bruno Haible)
2023 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2024 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2025 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2026 * bug fix: pathnames with relative directory components can be
2027 represented relative to default pathnames.
2028 * optimization: in taking the GCD of bignums, reduce the two bignums
2029 to approximately the same size (using Euclid's algorithm) before
2030 applying the more sophisticated binary GCD. (thanks to Juho
2032 * optimization: COUNT on bitvectors now operates word-at-a-time.
2033 * optimization: ASH with a positive, but not necessarily constant,
2034 (leftwards) shift, when in a modular context, is compiled to a
2036 * fixed some bugs revealed by Paul Dietz' test suite:
2037 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2038 if the corresponding argument is NIL.
2039 ** FORMAT directives accepting a minpad parameter treat negative
2041 ** literal commas as character parameters to FORMAT directives are
2043 ** literal spaces directly after ~<Newline> directives within a
2044 format-logical-block (~:< ~@:>) do not induce :FILL-style
2045 conditional newlines.
2046 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2047 as specified: it no longer includes conditional newlines).
2048 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2050 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2051 :PER-LINE-PREFIX argument does not evaluate to a string.
2053 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2054 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2055 the manual for details; for now, package locks can be disabled by
2056 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2057 if no major problems are found then it is likely that they will be
2058 compiled in unconditionally.
2059 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2060 unsupported operators on all platforms. To load a shared library
2061 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2062 object file, link it into a shared library outside of SBCL and
2063 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2065 * bug workaround: ROOM T and the GC were not working together
2066 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2067 reporting facility aren't preserved. That reporting has been
2068 disabled, so now until and unless someone figures out how to make it
2069 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2070 * minor incompatible change: as threatened around sbcl-0.8.0, the
2071 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2072 The symbols are also exported from SB-PCL for backwards
2073 compatibility, but more so than before SB-PCL should be treated as
2074 an implementation-internal package.
2075 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2077 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2078 It is known as of this release not to work on the Alpha, however.
2079 * fixed bug #167: errors signalled due to illegal syntax in method
2080 bodies are now more legible.
2081 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2082 designators and can hence be used with TYPEP.
2083 * fixed bug #333: CHECK-TYPE now ensures that the type error
2084 signalled, if any, has the right object to be accessed by
2085 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2086 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2087 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2089 * fixed bug #334: programmatic addition of slots using specialized
2090 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2091 and :ALLOCATION :CLASS slots.
2092 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2093 float exponents. (rereported by Peter Seibel)
2094 * fixed a bug: #\Space (and other whitespace characters) are no
2095 longer considered to be macro characters in standard syntax by
2096 GET-MACRO-CHARACTER.
2097 * fixed bug: initialization of condition class metaobjects no longer
2098 causes an instance of the condition to be created. (reported by
2100 * fixed bug: it is now possible to have slots such that
2101 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2102 description is neither :INSTANCE nor :CLASS.
2103 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2104 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2105 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2106 non-standard SLOT-DEFINITION-ALLOCATION values.
2107 * fixed bugs: various race conditions handling exiting threads.
2108 CL-PPCRE's thread torture test now passes.
2109 * fixed arguably-a-bug: GC time proportional to number of threads in
2110 system even when most of them are idle
2111 * optimization: improved performance of BIT and SBIT on bit-vectors.
2112 * .fasl file incompatibility: The fasl file version number has
2113 been incremented because of changes associated with package locks.
2115 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2116 * minor incompatible change: the system no longer provides
2117 optimization hints (conditions which are TYPEP
2118 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2119 mode; these hints are still emitted when higher SPEED optimization
2121 * new contrib module: a sampling profiler (profiling by statistical
2122 sampling, rather than by instrumenting functions) is available as
2123 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2124 * the behaviour of the standard function ED is now customizeable by
2125 third parties through a hook variable: see ED's documentation
2126 string for information on the protocol.
2127 * the compiler no longer emits efficiency notes for (FUNCALL X)
2128 when the type of X is uncertain under default optimization
2130 * fixed bug 276: mutating a binding of a specialized parameter to a
2131 method to something that is not TYPEP the specializer is now
2133 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2134 exhibit the required behaviour on the x86 platform. (thanks to
2135 Peter van Eynde, Eric Marsden and Bruno Haible)
2136 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2137 even for difficult arguments. (reported by Peter Graves)
2138 * fixed bug 141a: the backquote printer now descends quoted
2140 * fixed another bug in backquote printing: no more destructive
2141 modification of the form's list structure. (reported by Brian
2143 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2144 move between its address being taken and the call to
2145 interrupt_thread, fixing a crashing race condition.
2146 * the SB-POSIX contrib implementation has been adjusted so that it
2147 no longer exhibits ridiculously poor performance when constructing
2148 instances corresponding to C structs.
2150 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2151 * minor incompatible change: the sb-grovel contrib now treats C
2152 structures as alien (in the sense of SB-ALIEN) objects rather than
2153 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2154 has implications for memory management of client code
2155 (sb-grovel-returned objects must now be manually managed) and for
2156 type safety (alien objects now have full types).
2157 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2158 used to control emission of compiler diagnostics, rather than the
2159 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2160 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2161 quality should be considered deprecated.
2162 * install.sh now installs the user manual as well
2163 * (not quite a new documentable feature, but worth considering in
2164 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2165 beginnings of a semantically meaningful condition hierarchy is
2166 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2168 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2169 designator as the defaults argument.
2170 * fixed bug: Displaced arrays whose displaced-to array has become
2171 too small now cause ARRAY-DIMENSION to signal an error, providing
2172 for safer bounds-checking. (reported by Bruno Haible)
2173 * fixed bug: DEFCLASS slot definitions with identical :READER and
2174 :WRITER names now signal a reasonable error. (reported by Thomas
2176 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2178 * fixed bug: Invalid slot specification errors now print correctly.
2179 (thanks to Zach Beane)
2180 * fixed bug 320: Shared to local slot value transfers in class
2181 redefinitions now happen corrently from superclasses as
2182 well. (reported by Bruno Haible)
2183 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2185 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2186 improper lists correctly. (reported by Bruno Haible)
2187 * fixed bug 313: source-transform for <fun-name> was erroneously
2188 applied to a call of a value of a variable with name <fun-name>.
2189 (reported by Antonio Menezes Leitao)
2190 * fixed bug 307: The obsolete instance protocol ensures that
2191 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2192 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2193 remove dead unknown-values globs from the middle of the stack.
2194 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2195 easier to resume long computations after using *BREAK-ON-SIGNALS*
2196 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2197 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2198 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2199 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2200 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2201 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2203 * fixed a (fairly theoretical) bug in string printing: if
2204 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2205 does not have array-element-type equal to the most general string
2207 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2208 function name. (thanks to Nikodemus Siivola)
2209 * fixed bug: compile-time format string checker failed on
2210 non-closed ~{. (reported by Thomas F Burdick)
2211 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2212 don't warn on legal loop code involving more than one aggregate
2213 boolean. (thanks to Nikodemus Siivola)
2214 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2215 in keyword position to MAKE-PACKAGE was not being reported as
2217 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2218 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2220 * fixed bugs 280 and 312: the checking for multiple definitions in a
2221 file is less likely to become confused by uses of inline
2223 * fixed bug: the #S reader macro performs the keyword coercion
2224 specified for slot names. (reported by Kalle Niemitalo)
2225 * fixed bug: lambda lists may contain symbols whose names start with
2226 & but are not lambda-list-keywords; their occurrence triggers a
2228 * fixed bug 321: define-method-combination argument lambda lists do
2229 not cause invalid code to be generated when &optional or &aux
2230 variables have default values. (reported by Bruno Haible)
2231 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2232 have CLOS classes; this appears also to have fixed reported
2233 instability in stack exhaustion detection.
2234 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2235 MUFFLE-WARNING when no associated restart is present is now
2237 * optimization: rearranged the expansion of various defining macros
2238 so that each expands into only one top-level form in a
2239 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2241 * optimization: used a previously little-used slot in symbols to
2242 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2244 * fixed some bugs revealed by Paul Dietz' test suite:
2245 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2247 ** Fixed an optimization bug related to inheritance of initforms
2248 from local to shared slots.
2249 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2250 ** CAST optimizer forgot to flush argument derived type.
2251 ** print/read consistency on floats is now orders of magnitude
2252 more likely. (thanks also to Bruno Haible for a similar report
2254 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2255 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2256 for objects of type REAL. Make it so.
2257 ** SXHASH is specified (infelicitously) to respect similarity,
2258 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2259 it so. (thanks to Markus Ziegler)
2260 ** on the Alpha, the compiler succeeds in compiling functions
2261 returning a known number of arguments greater than 63.
2262 ** fixed handling of invalid NIL arguments in keyword position
2264 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2266 ** provided a readably-printable representation for RANDOM-STATE
2268 ** ensured that pathnames, where they have a namestring, always
2269 print using #P"..." syntax.
2271 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2272 * Support for the forthcoming 2.0 version of the NetBSD kernel
2273 running on x86 hardware has been added. (thanks to Perry
2274 E. Metzger most immediately, and others for their past work)
2275 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2276 current release version; SBCL's OpenBSD support had been broken
2277 since about the time of OpenBSD's switch to ELF binary format.)
2278 * [placeholder for DX summary]
2279 ** user code with &REST lists declared dynamic-extent, under high
2280 speed or space and low safety and debug optimization policy.
2281 * The manual has been converted to Texinfo format and the debugger
2282 chapter from the cmucl manual has been added.
2283 * A facility has been added to extract documentation strings from
2284 sbcl and store them as Texinfo-formatted snippets for inclusion in
2285 the manual (via Texinfo's @include directive)
2286 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2287 the test case to Dave Roberts)
2288 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2289 pointers. (reported by Sean Ross)
2290 * bug fix: PROFILE output is printed nicely even for large numerical
2291 values. (thanks to Zach Beane)
2292 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2293 greater than 32 handle EOF correctly.
2294 * bug fix: on X86 an immediate argument of the IMUL instruction is
2295 correctly printed in disassembly. (thanks to Lutz Euler)
2296 * bug fix: class slots in redefined classes preserve their old
2297 values. (thanks to Bruno Haible and Nikodemus Siivola)
2298 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2299 succeeds. (reported by Marco Baringer)
2300 * fixed some bugs revealed by Paul Dietz' test suite:
2301 ** READ-SEQUENCE now works on ECHO-STREAMs.
2302 ** RATIONALIZE works more according to its specification. (thanks
2305 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2306 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2307 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2308 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2309 should matter to you only if you rebind the printer control
2310 variables and then find you want different bindings in the
2311 debugger than in the ordinary execution of your program.)
2312 * The runtime build system has been tweaked to support building
2313 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2314 assembler and linker. (thanks to Nikodemus Siivola)
2315 * Unbound, undefined, undeclared variables now trigger full
2316 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2317 more often programmer error than deliberate exploitation of undefined
2319 * optimization: the hash algorithm for strings has changed to one
2320 that is less vulnerable to spurious collisions. (thanks to Juho
2322 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2323 needless bounds checking. (thanks to Juho Snellman)
2324 * optimization: implemented multiplication as a modular
2325 (UNSIGNED-BYTE 32) operation on the PPC backend.
2326 * fixed some bugs revealed by Paul Dietz' test suite:
2327 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2329 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2330 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2331 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2333 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2335 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2337 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2338 their output stream on EOF from read.
2339 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2340 have been read to end-of-file.
2341 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2343 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2344 description of determination of which consecutive characters
2346 ** Printing the "Space" character with escaping on now yields "#\\ ",
2347 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2348 ** Reading floating-point numbers with *READ-BASE* set to a number
2349 less than 10 works correctly.
2350 ** Reading floating-point numbers with *READ-BASE* set to a number
2351 more than 10 works correctly.
2352 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2353 the readtable currently in effect.
2355 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2356 * minor incompatible change: parsing of namestrings on a physical
2357 (Unix) host has changed; numbers after the final #\. in a
2358 namestring are no longer interpreted as a version field. This is
2359 intented to be largely invisible to the user, except that the
2360 meaning of the namestring "*.*.*" has changed: it now refers to a
2361 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2362 should usually be replaced by
2363 (make-pathname :name :wild :type :wild :version :wild)
2364 with the added benefit that this is more likely to be portable.
2365 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2366 option to OPEN now signals an error if the file being opened
2367 exists; this may have an impact on existing code.
2368 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2369 signals received, on the PowerPC platforms (both Linux and
2370 Darwin). (thanks to Pierre Mai for pointing out the location of
2372 * several fixes on OS X: The system now builds and runs cleanly on
2373 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2374 (thanks to Brian Mastenbrook)
2375 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2376 second-resolution: integer multiples of 1/3600 between -24 and 24.
2377 (thanks to Vincent Arkesteijn)
2378 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2379 type when called with 1 argument; PEEK-CHAR checked type of
2380 PEEK-TYPE only after having read first character from a
2381 stream. (reported by Peter Graves)
2382 * bug fix: the garbage collector now has much better locality
2383 behaviour, and in particular no longer treats all memory as being
2384 exhausted when many small objects point to each other in a deeply
2386 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2387 recognized as being TYPEP their class.
2388 * bug fix: the PUSHNEW documentation string has been corrected.
2389 (thanks to Vincent Arkesteijn)
2390 * bug fix: defaulting of the value for the last of an atypically
2391 large number of multiple values being bound was not being
2392 performed correctly on the Alpha or PPC platforms
2393 * optimization: implemented multiplication as a modular
2394 (UNSIGNED-BYTE 32) operation on the x86 backend.
2395 * optimization: SEARCH on simple-base-strings can now be open-coded.
2396 (see also contrib/compiler-extras.lisp for inspiration for
2397 teaching the compiler about the Boyer-Moore algorithm).
2398 * value, returned by MAX (and MIN) called with several EQUALP, but
2399 not EQL, arguments now does not depend on compiler settings.
2400 * fixed some bugs revealed by Paul Dietz' test suite:
2401 ** in stack analysis liveness information is propagated from
2402 non-local entry points.
2403 ** pathwise CAST removing failed when the CAST node did not start
2405 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2406 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2408 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2409 host is already defined.
2410 ** RENAME-FILE works on streams instead of signalling an internal
2412 ** PEEK-CHAR uses the current readtable when determining whether
2413 or not a character is whitespace.
2414 ** MERGE-PATHNAMES handles the case when the pathname does not
2415 specify a name while the default-pathname specifies a version.
2416 ** Pathnames now stand a better chance of respecting print/read
2418 ** Attempting to use standardized file system operators with a
2419 pathname with invalid :DIRECTORY components signals a
2421 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2423 ** DIRECTORY on logical pathnames is more correct.
2424 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2425 signal a TYPE-ERROR if their argument is not a stream.
2426 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2427 designator argument does not designate a stream.
2428 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2429 examining the synonym.
2430 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2432 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2433 element-type larger than ([UN]SIGNED-BYTE 32).
2435 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2436 * When built with the :SB-FUTEX feature, threaded builds now take
2437 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2438 for faster/more reliable mutex and condition variable support.
2439 * Incompatible change (but one you probably shouldn't have been using
2440 anyway): the interface and code for arbitrating between multiple
2441 threads in the same user session has been redesigned.
2442 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2443 error when called without an explicit environment argument.
2444 (thanks to Nikodemus Siivola)
2445 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2446 confused about their position. (thanks to Adam Warner and Gerd
2448 * bug fix: Pretty printing backquoted forms with unquotations in the
2449 argument list position of various code constructs such as LAMBDA
2450 now works correctly. (reported by Paul Dietz)
2451 * bug fix: Pretty printing unquotations no longer loses all
2452 stream position information.
2453 * optimization: performance of string output streams is now less
2454 poor for multiple small sequence writes.
2455 * optimization: performance of CSUBTYPEP in the presence of complex
2456 expressions involving CONS and NOT many times has been improved.
2457 (reported by Paul Dietz)
2458 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2460 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2461 streams for the REPL, for the debugger, and so on. (thanks to
2463 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2464 * fixed some bugs revealed by Paul Dietz' test suite:
2465 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2466 is ignored if it is too small, rather than propagating through
2467 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2468 ** extremely complex negations of CONS types were not being
2469 sufficiently canonicalized, leading to inconsistencies in
2471 ** VALUES tranformer lost derived type.
2473 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2474 * fixed a bootstrapping bug: the build process no longer assumes
2475 that the various BOOLE-related constants have the same value in
2476 host and target lisps. (noted by Paul Dietz' test suite on an
2477 SBCL binary built from CLISP)
2478 * The system can now be dynamically linked on the MIPS platform,
2479 which enables dynamic loading of foreign code from Lisp. (thanks
2480 to Ralf Baechle for discussions on the MIPS ABI)
2481 * The system now records debugging information for its own source
2482 files in a filesystem-position-independent manner, relative to
2483 the "SYS" logical host.
2484 * fixed a compiler bug: MV-LET convertion did not check references
2485 to the "max args" entry point. (reported by Brian Downing)
2486 * tweaked disassembly notes to be less confident about proclaiming
2487 some instruction as an LRA. (thanks to Brian Downing)
2488 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2489 now each have their own history, command character, and other
2490 characteristics. (thanks to David Lichteblau)
2491 * fixed some bugs revealed by Paul Dietz' test suite:
2492 ** compiler failure in compiling LOGAND expressions including a
2494 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2495 shift greater than 32.
2496 ** FLUSH-DEST did not mark blocks for type check regeneration.
2497 ** HANDLER-CASE failed to accept declarations in handler clauses
2498 in some circumstances.
2500 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2501 * New code in contrib/sb-introspect (still probably not entirely
2502 stable yet) provides some support for smart Lisp development
2503 environments like SLIME.
2504 * The conditions signalled for errors occurring when loading .fasl
2505 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2506 in a way which should help ASDF recover gracefully.
2507 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2508 stuff has been cleaned up. If you code contrib/ stuff, this might
2509 affect you, and you can look at contrib/README, contrib/STANDARDS,
2510 and/or the 0.8.4.27 diff to check.
2511 * In full calls the compiler now does not generate checks for declared
2512 argument types for all arguments.
2513 * various threading fixes
2514 ** and some experimental patches which didn't make it into
2515 the main tree for this release, but which are shipped in
2516 contrib/experimental-thread.patch as a possible fix for some
2517 failures (deadlock, spinning...) in GC-intensive multithreaded
2519 * fixed PPC build problem (source code incompatibility of different
2520 library versions): added offsetof() hackery which attempts to divine
2521 where glibc maintainers put uc_mcontext today
2522 * fixed bug 282: compiler does not trust type assertions while passing
2523 arguments to a full call.
2524 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2525 &OPTIONAL VALUES type parameter.
2526 * fix bug 214: algorithm for noting rejected templates is now more
2527 similar to that of template seletion. (also reported by rydis on
2529 * fixed bug 141b: printing backquoted information readably and prettily
2530 inserts a space where necessary.
2531 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2532 (CHARACTER 10) are now reported as errors, rather than propagated
2533 as unknown types. (reported by piso on #lisp)
2534 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2535 with values NIL and :ERROR. (thanks to Milan Zamazal)
2536 * fixed bug 191c: CLOS now does proper keyword argument checking as
2537 described in CLHS 7.6.5 and 7.6.5.1.
2538 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2539 longer raise an error; further, using a list as a for-as-arithmetic
2540 counter now raises a meaningful error.
2541 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2542 now understood by sequence creation functions such as MAKE-SEQUENCE
2544 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2545 a STRING-INPUT-STREAM.
2546 * compiler enhancement: SIGNUM is now better able to derive the type
2548 * type declarations inside WITH-SLOTS are checked. (reported by
2550 * fixed some bugs revealed by Paul Dietz' test suite:
2551 ** incorrect optimization of TRUNCATE for a positive first
2552 argument and negative second.
2553 ** compiler failure in let-convertion during flushing dead code.
2554 ** compiler failure while deriving type of TRUNCATE on an
2555 interval, containing 0.
2556 ** ASH of a negative bignum by a negative bignum count now returns
2558 ** intersection of CONS types now canonicalizes properly, fixing
2559 inconsistencies in SUBTYPEP.
2561 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2562 * incompatible change: The --disable-debugger command line
2563 option now clobbers the debugger at a more fundamental
2564 level, by redefining #'INVOKE-DEBUGGER instead of by
2565 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2566 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2567 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2568 enter the debugger REPL and then suffer recursive errors
2569 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2570 while under the new system, BREAK in --disable-debugger mode
2571 terminates the system just as an unhandled error would.
2572 * fixed compiler performance when processing loops with a step >1;
2573 * bug fix: DOCUMENTATION now retrieves generic function
2574 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2575 support has been systematized, and now supports the methods
2576 specified by ANSI, along with a default method and a method for
2577 slot documentation. (reported by Nathan Froyd)
2578 * bug fix: effective methods associated with a generic function are
2579 no longer cached over a change of that generic function's method
2580 combination. (reported by Andreas Fuchs)
2581 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2582 names a non-existent directory. (thanks to Andreas Fuchs)
2583 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2584 the PPC platform, lead to stack corruption; this has been fixed.
2585 (reported by Rainer Joswig)
2586 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2587 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2588 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2589 arithmetic in general, is now much more reliable on the PPC
2591 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2592 platform now returns the right answer.
2593 * optimization: restored some effective method precomputation in
2594 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2595 precomputation is now tunable.
2596 * optimization: compiler-internal data structure use has been
2597 reviewed, and changes have been made that should improve the
2598 performance of the compiler by about 20%.
2599 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2600 simple-base-strings and simple-bit-vectors is improved.
2601 * optimization: the optimization of 32-bit logical and arithmetic
2602 functions introduced in version 0.8.3 on the x86 has been
2603 implemented on the mips, ppc and sparc platforms; an
2604 implementation of the same facility, but for 64-bit arithmetic,
2605 has been added for the alpha.
2606 * microoptimization: the compiler is better able to make use of the
2607 x86 LEA instruction for multiplication by constants.
2608 * bug fix: in some situations compiler did not report usage of
2609 generic arithmetic in (SPEED 3) policy.
2610 * bug 145b fix: compiler used wrong type specifier while converting
2611 MEMBER-types to numeric.
2612 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2614 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2615 index argument now works properly on the Alpha platform.
2616 * bug fix: floating point exception treatment on the Alpha platform
2618 * bug fix: FILE-POSITION works much better on string input and
2619 output streams. (thanks to Nikodemus Siivola)
2620 * bug fix: many threading/garbage collection symptoms sorted.
2621 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2622 might be pseudo-atomic.
2623 * internal change: Stopping for GC is now done with signals not
2624 ptrace. GC is now done in whichever thread wanted it, instead of
2626 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2628 * bug fix: result form in DO is not contained in the implicit
2630 * incompatible change: ICR structure is changed; the value part of
2631 CONTINUATION is now called LVAR; corresponding functions are
2632 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2633 * added type deriver for ISQRT (thanks to Robert E. Brown).
2634 * bug fix: better support for loading from the command line when an
2635 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2637 * fixed some bugs revealed by Paul Dietz' test suite:
2638 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2639 ** ROUND and FROUND now give the right answer when given very
2640 small float arguments.
2641 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2643 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2644 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2645 ** LCM should return a non-negative integer.
2646 ** PARSE-INTEGER returned the index of a terminator instead of the
2647 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2648 ** PARSE-INTEGER returned an incorrect index being applied to a
2650 ** LCM with two arguments of 0 returns 0 rather than signalling
2652 ** unsigned addition of a 32-bit constant with the high bit set no
2653 longer causes an internal compiler error on the x86.
2654 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2655 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2656 with negative last argument.
2657 ** byte specifiers with zero size and position no longer cause
2658 an error during type derivation.
2659 ** bignum multiplication on the Alpha platform now returns the
2661 * porting: The system now builds on SuSE AMD64, although it still
2662 generates a 32-bit binary.
2663 * .fasl file incompatibility: The fasl file version number has
2664 been incremented (because of the changes to internal compiler
2665 data structures referred to above).
2667 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2668 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2669 more accurately, on the Darwin kernel running on PowerPC hardware.
2670 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2671 * Compiler code deletion notes now signal a condition of type
2672 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2673 an associated MUFFLE-WARNING restart.
2674 * The compiler now performs limited argument count validation of
2675 constant format strings in FORMAT, and where appropriate in ERROR,
2676 CERROR and WARN. (thanks to Gerd Moellmann)
2677 * New ASDF-INSTALL contrib can be used for automatic download and
2678 installation of third-party Lisp code from CCLAN or other sites
2680 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2681 which forces another thread to execute a function supplied by the
2683 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2684 now accept and act upon their :ELEMENT-TYPE keyword argument.
2685 (reported by Martin Atzmueller, Edi Weitz)
2686 * bug fix: FILE-POSITION now accepts position designators up to
2687 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2688 is the greater. (thanks to Patrik Nordebo)
2689 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2690 Kalle Olavi Niemitalo)
2691 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2692 printable. (reported by Eric Marsden)
2693 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2694 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2695 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2696 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2697 (thanks to Stig Sandoe)
2698 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2699 UNREAD-CHAR now correctly includes the unread character in the
2700 target sequence. (thanks to Gerd Moellmann)
2701 * bug fix in threaded builds: the system can now be suspended and
2702 resumed by shell job control with minimal disruption.
2703 * bug fixes in times and timezones >2038AD
2704 * better handling of "where is GNU make?" problem in build scripts
2705 (thanks to Nikodemus Siivola)
2706 * new optimization: inside a named function any reference to a
2707 function with the same name is considered to be a self-reference;
2708 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2709 optimization quality.
2710 * new optimization on x86: logical functions and + now have
2711 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2712 used when the result is truncated to 32 bits.
2713 * VALUES declaration is partially enabled.
2714 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2715 * fixed some bugs revealed by Paul Dietz' test suite:
2716 ** The system now obeys the constraint imposed by
2717 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2718 types form a lattice under type intersection.
2719 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2720 ** ASSOC now ignores NIL elements in an alist.
2721 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2722 and (1+ MOST-POSITIVE-FIXNUM) answers.
2723 ** The addition of a method with invalid qualifiers to a generic
2724 function does not cause an error to be signalled immediately;
2725 a warning is signalled, and the error is generated only on
2726 calling the generic function.
2727 * changed .fasl file version number, in part to add type codes for
2728 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2729 obscure ANSI requirements
2731 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2732 * fixed bug 148: failure to inline-expand a local function left
2733 garbage, confusing the compiler.
2734 * fixed bugs 3cd: structure slot readers perform type check if the
2735 slot can have an invalid value (i.e. it is either not initialized
2736 or can be written with a less specific slot writer).
2737 * bug fix: the compiler now traps array references to elements off
2738 the end of an array; previously, the bounds checking in some
2739 circumstances could go off-by-one.
2740 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2742 * type declarations for array element types now obey the description
2743 on the CLHS page "Declaration TYPE", as per discussions on
2744 sbcl-help around 2003-05-08. This means that a declaration
2745 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2746 declaration, all references to BAR will be asserted or assumed
2747 (with THE, so dependent on compiler policy) to involve objects of
2748 type FOO. Note that no such declaration is implied in
2749 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2750 * declared types of functions from the "Conditions"
2751 chapter. (reported by Paul Dietz)
2752 * bug fix: CERROR accepts a function as its first argument.
2753 * bug fix: NTH an NTHCDR accept a bignum as index
2754 arguments. (reported by Adam Warner)
2755 * optimization: character compare routines now optimize comparing
2756 against a constant character. (reported by Gilbert Baumann)
2757 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2758 argument now works properly on the MIPS platform.
2759 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2760 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2761 declared to have a functional type, does not cause an error
2763 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2764 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2765 * fixed bug: if last continuation of a deleted block has a
2766 destination, this destination should be deleted too. (reported by
2768 * fixed a bug in the bootstrap process: the host compiler's values
2769 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2770 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2771 test case from Patrik Nordebo)
2772 * improved the ability of the disassembler on the PPC platform to
2773 provide helpful disassembly notes.
2774 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2775 the class in more cases than previously.
2776 * bug fix: FILE-POSITION now understands :START and :END for
2777 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2778 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2779 8) data. (thanks to David Lichteblau)
2780 * bug fix: it is possible to add a method to a generic function
2781 without lambda list.
2782 * bug fix: reader failed to signal END-OF-FILE inside an
2783 object representation. (reported by Nikodemus Siivola)
2784 * fixed some bugs revealed by Paul Dietz' test suite:
2785 ** LAST and [N]BUTLAST should accept a bignum.
2786 ** condition slot accessors are methods.
2787 ** (VECTOR NIL) is a subtype of STRING.
2789 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2790 * minor incompatible change: some nonsensical specialized lambda
2791 lists (used in DEFMETHOD) which were previously ignored now signal
2793 * minor incompatible change: the system is now aware of the types of
2794 variables in the COMMON-LISP package, and will signal errors for
2795 most violations of these type constraints (where previously they
2796 were silently accepted).
2797 * minor incompatible change: COMPILE-FILE now uses the freedom
2798 afforded (ANSI 3.2.2.3) to use derived function types for
2799 functions defined in the same file. This also permits the system
2800 to warn on static type mismatches and function
2801 redefinition. (Currently it does not work with high DEBUG level.)
2802 * minor incompatible change: VALUES declaration is disabled.
2803 * When issuing notes, the compiler now signals a condition of type
2804 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2805 restart for use in user handlers. It is expected that the
2806 COMPILER-NOTE condition will eventually become a condition
2807 supertype to a hierarchy of note types, which will then be
2808 handleable in a similar fashion. However, at the moment, no such
2809 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2810 but it's an implementation detail, not a classification for the
2812 * Changes in type checking closed the following bugs:
2813 ** type checking of unused values (192b, 194d, 203);
2814 ** template selection based on unsafe type assertions (192c, 236);
2815 ** type checking in branches (194bc).
2816 * A short form of VALUES type specifier has ANSI meaning (it has
2817 increased the number of situations when SBCL cannot perform type
2819 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2820 keywords or constants is permissible.
2821 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2822 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2823 classes. (thanks to Antonio Martinez)
2824 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2825 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2826 argument to a pretty-print stream. Instead, it leaves any such
2827 operations to DESCRIBE-OBJECT methods.
2828 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2830 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2831 lambda lists. (thanks to David Lichteblau)
2832 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2833 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2834 (thanks to Gerd Moellmann)
2835 * fixed bug 246: increased compilation speed of long
2836 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2838 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2840 * DEFCONSTANT now throws a condition of type
2841 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2842 constant to a non-EQL value; CONTINUE and ABORT restarts
2843 respectively change and preserve the value.
2844 * fixed bug 63: The code walker, part of the implementation of CLOS,
2845 is now better at handling symbol macros.
2846 * bug fix: There is no longer an internal implementation type named
2847 CL:LENGTH. (reported by Raymond Toy)
2848 * bug fix: In macro-like defining macros/special operators the
2849 implicit block does not enclose the lambda list.
2850 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2851 enclosed in parentheses) are not suitable as type specifiers, and
2852 their use properly signals an error now.
2853 * bug fix: An argument count mismatch for a type specifier in code
2854 being compiled no longer causes an unhandled error at compile
2855 time, but signals a compile-time warning.
2856 * fixed simple vector readable printing
2857 * bug fix: DESCRIBE takes more care over whether the class
2858 precedence list slot of a class is bound before accessing it.
2859 (reported by Markus Krummenacker)
2860 * bug fix: FORMATTER can successfully compile pretty-printer format
2861 strings which use variants of the ~* directive inside.
2862 * bug fix: SEARCH now applies its TEST predicate to the elements of
2863 the arguments in the correct order. (thanks to Wolfhard Buss)
2864 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2865 in type inference. (thanks to Robert E. Brown)
2866 * bug fix: Reading in symbols with an explicit package name of ""
2867 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2868 * fixed some bugs revealed by Paul Dietz' test suite:
2869 ** NIL is now allowed as a structure slot name.
2870 ** Arbitrary numbers, not just REALs, are allowed in certain
2871 circumstances in LOOP for-as-arithmetic clauses.
2872 ** Multiple class redefinitions before slot access no longer
2873 causes a type error.
2874 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2875 association between the name and a class.
2876 ** Generic functions with non-standard method-combination and over
2877 five methods all of which return constants no longer return NIL
2878 after the first few invocations. (thanks to Gerd Moellmann)
2879 ** CALL-NEXT-METHOD with no arguments now passes the original
2880 values of the arguments, even in the presence of assignment.
2881 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2883 ** DISASSEMBLE works with closures and funcallable instances.
2884 ** ADD-METHOD now returns the generic function, not the new method.
2885 ** FIND-METHOD signals an error if the lengths of the specializers
2886 is incompatible with the generic function, even if the ERRORP
2888 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2889 which its argument is a member.
2890 ** DEFCLASS only redefines the class named by its class-name
2891 argument if that name is the proper name of the class;
2892 otherwise, it creates a new class.
2893 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2894 of the UNBOUND-SLOT condition to the name of the slot.
2895 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2896 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2897 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2898 SLOT-MAKUNBOUND in the specified fashion.
2900 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2901 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2902 cross-compilation host. As a consequence, we can now bootstrap our
2903 way up to SBCL starting with a bare gcc toolchain and human-readable
2904 source code (first the source to CLISP, then the source to SBCL).
2905 * A contributed module containing a partial implementation of the
2906 simple-streams interface has been included. (thanks to Rudi
2908 * A contributed module implementing the RFC1321 Message Digest
2909 Algorithm, known as MD5, has been included.
2910 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2911 no longer has any effect, as the code controlled by this feature
2912 has been deleted. (As far as we know, no-one has ever built using
2913 this feature, and its semantics were confused in any case).
2914 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2915 work on conditions (as required by the ANSI specification),
2916 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2917 expected behaviour on conditions. Users should note, however,
2918 that such behaviour is not required by the ANSI specification,
2919 and so use of this behaviour may render their code unportable.
2920 * fixed some bugs revealed by Paul Dietz' test suite:
2921 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2923 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2924 ** SXHASH on condition objects no longer returns NIL.
2925 ** :ALLOCATION :CLASS slots are better treated; their values are
2926 updated on class redefinition, and initforms inherited from
2927 superclasses are applied.
2928 ** REMOVE-METHOD returns its generic function argument even when
2929 no method was removed.
2930 ** SHARED-INITIALIZE now initializes the values of the requested
2931 slots, including those with :ALLOCATION :CLASS.
2932 ** ALLOCATE-INSTANCE now works on structure classes defined via
2933 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2935 ** SLOT-EXISTS-P now works on conditions, as well as structures
2937 ** MAKE-LOAD-FORM now has the required methods on
2938 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2939 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2940 keyword, and now implements the SLOT-NAMES argument.
2941 ** methods with &OPTIONAL arguments no longer allow too many
2942 arguments to be passed in the call without error.
2943 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2944 option is consistent with the required arguments of the generic
2945 function lambda list.
2946 * bug fix: REQUIRE accepts a string designator. (Thanks to
2948 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2949 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2950 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2951 * bug fix: The compiler now checks for duplicated variables in macro
2953 * bug fix: SETQ on globals returns the correct value.
2954 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2955 not a COMPILER-ERROR (followed by some other strange error on
2956 choosing the CONTINUE restart).
2957 * bug fix: make.sh and friends are now more consistent in the way that
2958 they look for GNU "make".
2960 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2961 * experimental native threads support (on x86 Linux >=2.4 only).
2962 This is not compiled in by default: you need to add :SB-THREAD to
2963 the target features. See the "Beyond ANSI" chapter of the manual
2965 * fix for longstanding nonANSIism: The old distinction between
2966 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2967 The return value from CL:FIND-CLASS is now a CLOS class, and
2968 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2969 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2970 * An interface to the MetaObject Protocol, as described in Kiczales,
2971 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2972 MIT Press, 1991, is available from the SB-MOP package.
2973 * incompatible change: the SB-PCL package should now be considered
2974 a private implementation detail, and no longer a semi-private MOP
2976 * minor incompatible change: due to rearrangement for threads, the
2977 control stack and binding stack are now allocated at arbitrary
2978 addresses instead of being hardcoded per-port. Users affected by
2979 this probably have to be doing advanced things with shared
2980 libraries, and will know who they are.
2981 * minor incompatible change: Previously, all --eval forms used were
2982 processed with READ before any of them were processed with EVAL.
2983 Now each --eval form is processed with both READ and EVAL before
2984 the next --eval form is processed. (Thus package operations like
2985 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2986 work as the user might reasonably expect.)
2987 * minor incompatible change: *STANDARD-INPUT* is now only an
2988 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2990 * minor incompatible change: Y-OR-N-P is now character-oriented,
2991 not line oriented. Also, YES-OR-NO-P now works without errors.
2992 (thanks to Antonio Martinez)
2993 * sb-aclrepl module improvements: an integrated inspector, added
2994 repl features, and a bug fix to :trace command.
2995 * Known functions, which cannot be open coded by the backend, are
2996 considered to be able to check types of their arguments. (fixing
2997 a bug report by Nathan J. Froyd)
2998 * fixed a bug in computing method discriminating functions: It is
2999 now possible to define methods specialized on classes which have
3000 forward-referenced superclasses. (thanks to Gerd Moellmann)
3001 * fixed evaluation order in optional entries (reported by Gilbert
3003 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3004 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3005 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3006 required-by-AMOP NAME argument, as well as CLASS and
3007 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3008 * fixed bug 20: DEFMETHOD can define methods using names that are
3009 not the proper names of classes to designate class specializers.
3010 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3011 Unix file descriptors, instead of blowing up. (thanks to Antonio
3013 * Garbage collection refactoring: user-visible change is that a
3014 call to the GC function during WITHOUT-GCING will not do garbage
3015 collection until the end of the WITHOUT-GCING. If you were doing
3016 this you were probably losing anyway.
3017 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3018 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3019 TYPEP the latter but not the former.
3020 * The compiler issues a full WARNING for calls to undefined functions
3021 with names from the CL package.
3022 * MAP-INTO for a vector destination is open coded. (reported by
3023 Brian Downing on c.l.l)
3024 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3025 documentation string.
3026 * fixed some bugs revealed by Paul Dietz' test suite:
3027 ** COPY-ALIST now signals an error if its argument is a dotted
3029 ** Condition slots are now accessed more correctly in the presence
3030 of multiple initargs for a given slot.
3031 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3032 exclude restarts of the same name associated with a different
3034 ** DEFCLASS of forward-referenced classes with another
3035 forward-referenced class in the superclasses list no longer
3037 ** Condition slots are now initialized once each, not multiple
3038 times. (thanks to Gerd Moellmann)
3039 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3041 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3042 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3043 arguments contain duplicated elements.
3044 ** RESTART-CASE understands local macros.
3045 ** RESTART-CASE associates exactly its own restarts with a condition.
3046 ** ENDP in safe mode checks its argument to be of type LIST.
3047 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3048 in question is unbound.
3049 ** Optimized MAKE-INSTANCE functions no longer cause internal
3050 assertion failures in the presence of duplicate initargs.
3051 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3052 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3053 propagated even in the ctor optimized implementation of
3055 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3057 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3058 * a better implementation of SXHASH on (simple) bit vectors,
3059 measured both in execution speed and in distribution of results
3060 over the positive fixnums, has been installed. Likewise, a better
3061 implementation of EQUAL for simple bit vectors is now available.
3062 * fixed CEILING optimization for a divisor of form 2^k.
3063 * fixed bug 240 (emitting extra style warnings "using the lexical
3064 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3065 by Antonio Martinez)
3066 * fixed SXHASH, giving different results for NIL depending on type
3067 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3068 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3069 variable. (found by Rolf Wester)
3070 * SBCL does not ignore type declarations for special
3071 variables. (reported by rif on c.l.l 2003-03-05)
3072 * some bug fixes in contrib/sb-aclrepl/
3073 * fixed some bugs revealed by Paul Dietz' test suite:
3074 ** a bug in the CONS type specifier, whereby the CAR and CDR
3075 types got intertwined, has been fixed;
3076 ** the type system is now able to reason about the interaction
3077 between INTEGER and RATIO types more completely;
3078 ** APPEND, [N]REVERSE and NRECONC check that those their
3079 arguments, which must be proper lists, are really so;
3080 ** An array specialized to be unable to hold elements has been
3081 implemented, as required -- yes, really -- by ANSI;
3082 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3083 on malformed property lists;
3085 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3086 * incompatible packaging change: in line with Unix convention,
3087 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3088 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3089 * REQUIRE and PROVIDE are now optionally capable of doing something
3090 useful. See the documentation string for REQUIRE.
3091 * infrastructure for a managed SBCL contrib system: contributed
3092 modules in this release include:
3093 ** the ASDF system definition facility;
3094 ** an interface to the BSD Sockets API;
3095 ** an ACL-like convenience interface to the repl;
3096 (thanks to Kevin Rosenberg)
3097 ** an implementation of ROTATE-BYTE, with efficient implementation
3099 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3100 gives a closer approximation to the right answer than previously.
3101 (thanks to Raymond Toy)
3102 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3103 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3104 optional environment arguments, as required by ANSI.
3105 * fixed bugs in other functions taking environment objects, allowing
3106 calls with an explicit NIL environment argument to be compiled
3108 * fixed bug 228: primary return values from
3109 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3110 COMPILE or FUNCTION.
3111 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3112 :TYPE will no longer signal errors on innocuous objects.
3113 * fixed bug 231b: SETQ is better at respecting type declarations in
3114 the lexical environment.
3115 * fixed a bug in DEFCLASS: classes named by symbols with no or
3116 unprintable packages can now be defined.
3117 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3118 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3119 * fixed compiler failure related to checking types of functions.
3120 (reported by Robert E. Brown)
3121 * the compiler is now much more consistent in its error-checking
3122 treatment of bounding index arguments to sequence functions: in
3123 (SAFETY 3) code, errors will be signalled in almost all cases if
3124 invalid sequence bounding indices are passed to functions defined
3125 by ANSI to operate on sequences.
3126 * fixed a bug in the build procedure: documentation of SBCL-specific
3127 packages is now preserved and available in the final Lisp image.
3128 * lifted FDEFINITION lookup out of loops in the implementation of
3129 many list operations. (thanks to Robert E. Brown)
3130 * fixed a bug in the reader: the #n# reader macro now works for
3131 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3132 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3133 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3134 E. Brown and Tony Martinez respectively)
3135 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3136 SLOT-BOUNDP: in particular, it is now able to optimize them much
3137 better, and is now not vulnerable to having packages renamed.
3138 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3139 always called when a slot is not present in an instance. (thanks
3141 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3142 longer signal an error on unreading a character following EOF on
3143 the previous constituent stream. (thanks to Tony Martinez)
3144 * fixed some bugs revealed by Paul Dietz' test suite:
3145 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3146 not just nonnegative fixnums;
3147 ** the logical bit-array operators such as BIT-AND now accept an
3148 explicit NIL for their "opt-arg" argument (to indicate a
3149 freshly-consed result bit-array);
3150 ** ELT now signals an error on an invalid sequence index in safe
3152 ** the type system is now cleverer about negations of numeric
3153 types, and consequently understands the BIGNUM and RATIO types
3155 ** the type system is now cleverer about the interaction between
3156 INTEGER and RATIO types: while bugs still remain, many more
3157 cases are accurately computed;
3158 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3159 if it is in the last clause;
3160 ** CONSTANTLY now correctly returns a side-effect-free function in
3162 ** DECLARE is no longer treated as a special-operator; in
3163 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3164 * incremented fasl file version number due to the change in the
3165 DEFSTRUCT-SLOT-DESCRIPTION structure.
3167 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3168 * minor incompatible change: code processed by the "interpreter" or
3169 EVAL now has a compilation optimization policy of (DEBUG 2)
3170 (changed from (DEBUG 1)) to improve debuggability of interactive
3171 development, and to allow the use of the debug RETURN command in
3173 * an experimental implementation of the RETURN command for the
3174 debugger has been included. (thanks to Frederik Kuivinen)
3175 * fixed bug 62: constraints were not propagated into a loop.
3176 * fixed bug in embedded calls of SORT (reported and investigated by
3178 * fixed some bugs revealed by Paul Dietz' test suite:
3179 ** printing and reading of arrays with some dimensions having
3180 length 0 (thanks to Gerd Moellmann);
3181 ** BOA constructor with &AUX argument without a default value does
3182 not cause a type error;
3183 ** CONSTANTP now returns true for all self-evaluating objects.
3185 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3186 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3187 accessors that are related by inheritance, as specified in the
3188 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3189 Valtteri Vuorikoski)
3190 * The compiler is now able to inline functions that were defined in
3191 a complex lexical environment (e.g. inside a MACROLET).
3192 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3194 * Support for the upcoming FreeBSD-5.0 release has been included.
3195 (thanks to Dag-Erling Smorgrav)
3196 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3197 effect when it is not in a toplevel context.
3198 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3199 stand a better chance of being correct. (thanks to Gerd
3201 * fixed bug in COERCE, which now signals an error on coercing a
3202 rational to a bounded real type which excludes the expected
3204 * The compiler is now able to derive types more accurately from the
3205 COERCE and COMPILE functions.
3206 * fixed bug 223: functional binding is considered to be constant
3207 only for symbols in the CL package.
3208 * fixed bug 231: SETQ did not check the type of a variable being set
3209 (reported by Robert E. Brown)
3210 * A new optimization for MAKE-INSTANCE has been included, fixing
3211 various bugs (including relating to :ALLOCATION :CLASS slots and
3212 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3214 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3215 ** As required by ANSI, LOOP now disallows anonymous collection
3216 clauses such as COLLECT I in conjunction with aggregate boolean
3217 clauses such as THEREIS (= I 1);
3218 ** LOOP now signals an error when any variable is reused in the
3219 same loop (including the potentially useful construct analogous
3220 to WITH A = 1 WITH A = (1+ A);
3221 ** IT is only a special loop symbol within the first clause of a
3222 conditional loop clause;
3223 ** LOOP with a typed iteration variable over a hashtable now
3224 signals a type error iff it should.
3225 * fixed some other bugs revealed by Paul Dietz' test suite:
3226 ** FILE-STREAM now names the class previously known as FD-STREAM;
3227 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3228 argument) no longer signals an error;
3229 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3230 of the slot symbol, rather than using the current package
3231 ((:CONC-NAME "") continues to intern the slot's name in the
3233 * incremented fasl file version number, because of the incompatible
3234 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3235 of the new implementation of DEFINE-COMPILER-MACRO.
3237 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3238 * Support for building SBCL for MIPS platforms running in
3239 little-endian mode has now been checked in, and basic
3240 functionality on said platforms verified.
3241 * minor incompatible change: PCL now records the pathname of a file
3242 in which methods and the like are defined, rather than its
3244 * minor incompatible change: TRUENAME now considers the truename of
3245 a file naming a directory to be the pathname with :DIRECTORY
3246 component indicating that directory.
3247 * minor incompatible change: a NAMED clause in the extended form of
3248 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3249 reason for the previous behaviour is unclear.
3250 * more systematization and improvement of CLOS and MOP conformance
3251 in PCL (thanks to Gerd Moellman and Pierre Mai):
3252 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3254 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3255 primary methods with no specializers;
3256 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3258 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3259 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3260 FORWARD-REFERENCED-CLASSes; error reporting on
3261 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3263 ** SXHASH on CLOS instances now uses a slot internal to the
3264 instance to return different numbers on distinct instances,
3265 while preserving the same return value through invocations of
3267 ** DEFMETHOD signals errors when methods with longer incongruent
3268 lambda lists are added to generic functions;
3269 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3270 CLASS, as specified in AMOP;
3271 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3272 on the order returned by the primary method for classes of
3273 class STANDARD-CLASS;
3274 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3275 * fixed some bugs shown by Paul Dietz' test suite:
3276 ** DOLIST puts its body in TAGBODY;
3277 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3279 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3280 value producing form;
3281 ** if more variables are given to PROGV than values, extra
3282 variables are bound and made to have no value;
3283 ** NSUBSTITUTE on list arguments gets the right answer with
3285 ** ELT signals an error of type TYPE-ERROR when the index argument
3286 is not a valid sequence index;
3287 ** LOOP signals (at macroexpansion time) an error of type
3288 PROGRAM-ERROR when duplicate variable names are found;
3289 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3290 ** FUNCALL of special-operators now cause an error of type
3292 ** PSETQ now works as required in the presence of side-effecting
3293 symbol-macro places;
3294 ** NCONC accepts any object as its last argument;
3295 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3297 ** loop-for-as-package does not require a package to be explicitely
3299 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3301 * fixed bug 166: compiler preserves "there is a way to go"
3302 invariant when deleting code.
3303 * fixed bug 172: macro lambda lists with required arguments after
3304 &REST arguments now cause an error to be signalled. (thanks to
3306 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3307 bug. (thanks to Matthew Danish)
3308 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3310 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3311 arguments are lexically rebound. (thanks to Gerd Moellmann and
3313 * fixed bug 194: error messages are now more informative when there
3314 is no primary method applicable in a call to a generic
3315 function. (thanks to Gerd Moellmann)
3316 * fixed bug in command line argument checking (thanks to Julian
3318 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3319 (thanks to Matthew Danish)
3320 * incremented fasl file version number, because of the
3321 SXHASH-related changes in the layout of CLOS data structures
3323 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3324 * minor incompatible change: The runtime (the Unix executable named
3325 "sbcl") is now much pickier about the .core files it will load.
3326 Essentially it now requires .core files to descend from the same
3327 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3328 as the runtime does. (The intent is to prevent the crashes which
3329 can occur, and which can even be reported as mysterious failures,
3330 when people patch the sources or change the build parameters
3331 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3332 sbcl and .core files.)
3333 * fixed bug: VALUES-LIST is no longer optimized away.
3334 * fixed bug 142: The FFI conversion of C string values to Lisp
3335 string values no longer conses excessively. (thanks to Nathan
3336 Froyd porting Raymond Toy's fix to CMU CL)
3337 * began to systematize and improve MOP conformance in PCL (thanks to
3338 Nathan Froyd, Gerd Moellman and Pierre Mai):
3339 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3341 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3342 ** FINALIZE-INHERITANCE is now called on class finalization;
3343 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3344 argument precedence order.
3345 * fixed bug 202: The compiler no longer fails on functions whose
3346 derived types contradict their declared type.
3347 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3348 so it can be non-toplevel.
3349 * The fasl file version number has changed (because of the new
3350 implementation of DEFMACRO).
3351 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3352 safe code, that any length requirement by their type-specifier
3353 argument is valid. The exceptions to this are described in bug
3355 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3356 functions now check the types of their inputs as required by ANSI.
3357 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3358 introduced symbol is DECLAREd to be SPECIAL.
3359 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3360 * fixed bug: :COUNT argument to sequence functions may be negative
3361 * fixed bug: body of DO-SYMBOLS may contain declarations
3362 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3363 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3364 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3365 symbol macro only once
3366 * fixed printing of call frame when argument list is unavailable
3367 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3368 * compiler no longer signals WARNING on unknown keyword
3371 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3372 * A beta-quality port to the mips architecture running Linux,
3373 based on the old CMUCL backend, has been made. It has been tested
3374 on a big-endian kernel, and works sufficiently well to be able to
3375 rebuild itself; it has not been tested in little-endian mode.
3376 * fixed an inconsistency between gencgc.c and purify.c which made
3377 dumping/loading .core files unreliable
3378 * fixed bug 120a: The compiler now deals correctly with IFs where
3379 the consequent is the same as the alternative, instead of
3380 misderiving the return type. (thanks to Alexey Dejneka)
3381 * fixed bug 113: Logical pathnames are now dumpable (the logical
3382 host is resolved at load-time, throwing an error if it is not
3384 * fixed bug 174: FORMAT's error message is slightly clearer when a
3385 non-printing character is used in a format directive.
3386 * fixed several bugs in compiler checking of type declarations, i.e.
3387 violations of the Python "declarations are assertions" principle
3388 (thanks to Alexey Dejneka)
3389 * fixed several bugs in PCL's error checking (thanks to Gerd
3391 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3393 * fixed bug in compilation of functions as first class values
3394 (thanks to Antonio Martinez-Shotton)
3395 * The compiler's handling TYPE-ERRORs which it can prove will
3396 inevitably happen at runtime has been cleaned up and corrected
3397 in several ways. (thanks to Alexey Dejneka)
3398 * improved argument type checking for various basic arithmetic
3399 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3400 lavished on them in the past that they can be compiled in many
3401 ways in different special cases
3402 * fixed bug 181: compiler checks validity of user supplied type
3404 * cleaned up code flushing in optimization: Function calls which
3405 should signal errors for safety purposes (e.g. which ANSI says
3406 should signal errors when their arguments are of incorrect type)
3407 are no longer optimized away.
3408 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3409 * incremented fasl file version number, because changes in the
3410 implementation of sequence functions like COERCE caused
3411 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3412 in old inline expansions) to become undefined. (Actually these
3413 changes were later undone, so we might very well be binary
3414 compatible with 0.7.7 after all, but leaving the version number
3415 incremented seemed like the simplest and most conservative
3418 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3419 * An alpha-quality port to the parisc architecture running Linux,
3420 based on the old CMUCL backend, has been made. This, even more so
3421 than the other backends, should be considered still a work in
3422 progress; known problems include that the Linux kernel in 64-bit
3423 mode does not propagate the correct sigcontext structure to
3424 userspace, and consequently SBCL on a parisc64 kernel will not
3426 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3427 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3428 suppressed.) Also now that the compiler is looking at declarations
3429 in the environment, it checks optimization declarations as well,
3430 and suppresses inlining when (> DEBUG SPEED).
3431 * More fixes have been made to treatment of floating point exception
3432 treatment and other Unix signals. In particular, floating point
3433 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3434 * The detection and handling of control stack exhaustion (infinite
3435 or very deeply nested recursion) has changed. Stack exhaustion
3436 detection is now done by write-protecting pages at the OS level
3437 and applies at all optimization settings; when found, a
3438 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3439 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3440 be caught using IGNORE-ERRORS.
3441 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3442 correctly when transforming recursive LABELS functions to LETs.
3443 (thanks to Alexey Dejneka)
3444 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3445 that are names of constants or global variables.
3446 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3447 alien routines with docstrings.
3448 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3449 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3451 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3452 (thanks to Eric Marsden)
3453 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3454 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3455 * Bug fix: The SPARC backend can now compile functions involving
3456 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3457 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3458 to a functions expecting a PATHNAME-DESIGNATOR.
3459 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3460 lambda lists. (thanks to Alexey Dejneka)
3461 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3462 functions so that EXPORTing the name of the class doesn't cause
3463 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3464 to Antonio Martinez for reporting this)
3465 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3466 constant-folded without failing an assertion. (thanks to Einar
3467 Floystad Dorum for reporting this)
3468 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3469 types are now dealt with more correctly by the compiler.
3470 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3471 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3472 behaviour analogous to RENAME-FILE. This puts its behaviour more
3473 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3475 * The fasl file version number has changed again. (because of the
3476 bug fix involving the names of PCL MAKE-INSTANCE functions)
3478 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3479 * bug fix: Floating point exceptions are treated much more
3480 consistently on the x86/Linux and PPC/Linux platforms.
3481 * Array initialization with :INITIAL-ELEMENT is now much faster for
3482 cases when the compiler cannot open code the array creation, but
3483 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3484 array accesses have also seen a speed increase.
3485 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3486 specified. (This was at the root of some bad interactions between
3487 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3489 * bug fix: Internal error arguments for undefined functions are now
3490 computed correctly on the PPC/Linux platform.
3491 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3492 Raymond Toy's patch for CMU CL)
3493 * Support for the Solaris 9 operating environment has been included
3494 (thanks to Daniel Merritt)
3495 * A very ugly but hopefully complete draft of the missing FFI chapter
3496 of the manual has been created by reformatting the corresponding
3497 CMU CL manual chapter into (currently very ugly and incoherent)
3498 DocBook and bringing it up to date for SBCL behavior. Thus, the
3499 manual is now essentially complete, at least by my extreme
3500 once-and-only-once standards, whereby it's acceptable to refer to
3501 the doc strings of SB-EXT functions as the primary documentation.
3502 * The fasl file version number has changed again, due to cleanup of
3503 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3504 is no longer a static symbol.)
3506 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3507 * SBCL now builds with OpenMCL (version 0.12) as the
3508 cross-compilation host; also, more progress has been made toward
3509 bootstrapping under CLISP.
3510 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3512 * bug 158 fixed: The compiler can now deal with integer loop
3513 increments different from 1; fixing this turned out also to fix
3515 * bug 169 fixed: no more bogus warnings about using lexical bindings
3516 despite the presence of perfectly good SPECIAL declarations (thanks
3517 to David Lichteblau)
3518 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3519 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3520 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3522 * bug 180 fixed: Method combination specifications no longer ignore
3523 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3524 * bug fix: Structure type predicate functions now check their argument
3525 count as they should.
3526 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3527 correctly. (thanks to Pierre Mai)
3528 * minor incompatible change: The --noprogrammer option is deprecated
3529 in favor of the new --disable-debugger option, which is very similar.
3530 (The major difference is that it takes effect at a slightly different
3531 time at startup, causing handling of errors in --sysinit and
3532 --userinit files will be affected differently.) The
3533 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3534 been added to allow this functionality to be controlled from ordinary
3535 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3536 maintainers, who might want to run non-interactive scripts to
3537 build SBCL cores which will later be used interactively.)
3538 * minor incompatible change: The LOAD function no longer, when given
3539 a wild pathname to load, loads all files matching that pathname.
3540 Instead, an error of type FILE-ERROR is signalled.
3542 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3543 * bug 147 fixed: The compiler preserves its block link/count
3544 invariants more correctly now so that it doesn't crash. (thanks
3546 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3548 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3549 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3550 no-value-for-this-character as NIL (as specified by ANSI).
3551 * HOST-NAMESTRING on physical pathnames now returns a string that is
3552 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3553 (thanks to Christophe Rhodes)
3554 * The Alpha port handles icache flushing more correctly. (thanks to
3556 * More progress has been made toward bootstrapping under CLISP. (thanks
3557 to Christophe Rhodes)
3558 * The fasl file format has changed again, because dynamic loading
3559 on OpenBSD (which has non-ELF object files) motivated some cleanups
3560 in the way that foreign symbols are transformed and passed around.
3561 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3562 is no longer treated as whitespace by the reader, but instead as
3563 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3564 |A
\7fB|, instead of A as it used to.
3566 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3567 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3568 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3569 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3570 this as of 0.7.1.45, but was left out of the previous news section
3571 (thanks to Dan Barlow)
3572 * SBCL now runs on the Solaris operating system on SPARC architectures
3573 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3574 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3575 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3576 * SPARC backend cleanups, allowing builds of cores optimized for V8
3577 and V9 SPARCS, and also emission of code targeted to a particular
3578 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3580 * SBCL is closer to bootstrapping under CLISP, thanks to various
3581 fixes by Christophe Rhodes.
3582 * The fasl file format has changed again, to allow the compiler's
3583 INFO database to support symbol macros.
3584 * The user manual (in doc/) is formatted into HTML more nicely.
3585 (thanks to coreythomas)
3586 * The system is smarter about SUBTYPEP relationships, especially
3587 those involving NOT types (including types such as ATOM which are
3588 represented internally using NOT types). Thus SUBTYPEP is less
3589 likely to return (VALUES NIL NIL) in general, and in particular
3590 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3592 * The fasl file format has changed again, because the internal
3593 representation of types now includes a new slot to support the new
3594 SUBTYPEP-of-NOT-types logic.
3595 * (not a change in the main branch of SBCL, but a related prototype
3596 which can hopefully be merged into the main branch of SBCL in the
3597 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3598 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3599 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3600 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3601 dispatch macro character combinations. (thanks to Alexey Dejneka)
3602 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3603 that superficially look like logical namestrings correctly.
3604 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3606 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3607 default supplied-p) for &optional and &key arguments. (thanks to
3610 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3611 * incompatible change: The compiler is now less aggressive about
3612 tail call optimization, doing it only when (> SPACE DEBUG) or
3613 (> SPEED DEBUG). (This is an incompatible change because there are
3614 programs which relied on the old CMU-CL-style behavior to optimize
3615 away their unbounded recursion which will now die of stack overflow.)
3616 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3617 for non-GENCGC systems has been increased to 20M (since that
3618 seems much closer to the likely performance optimum for modern
3619 systems than the old 4M value was)
3620 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3621 and *DEBUG-PRINT-LEVEL*
3622 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3623 of CMU CL's support for SPARC, and various endianness and other
3624 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3625 * new syntactic sugar for the Unix command line: --load foo.bar is now
3626 an alternate notation for --eval '(load "foo.bar")'.
3628 ** The system now detects stack overflow and handles it gracefully,
3629 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3630 optimization settings. (This is a good thing in general, and
3631 its introduction in this version should be particularly timely
3632 for anyone whose code fails because of suppression of tail
3634 ** The system now hunts for the C variable "environ" in a more
3635 devious way, to avoid segfaults when the C library version differs
3636 between compile time and run time. (thanks to Christophe Rhodes)
3637 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3638 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3639 ** The compiler no longer issues bogus style warnings for undefined
3640 classes in the same source file as the DEFCLASSes which defined
3641 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3643 ** fixes in CONDITION class precedence list for undefined function
3644 errors (thanks to Alexei Dejneka)
3645 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3646 correctly. (thanks to Dan Barlow)
3647 ** portability fixes aiming at bootstrapping under CLISP (thanks
3648 to Dave McDonald and Christophe Rhodes)
3649 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3650 ** fixes in type translation and and type inference (thanks to
3652 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3653 ** various fixes in the new ports (thanks to Dan Barlow)
3654 * several changes related to debugging:
3655 ** suppression of tail recursion, as noted above
3656 ** stack overflow detection, as noted above
3657 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3658 is now the default. (For some time encapsulation has been more
3659 reliable than the breakpoint-based :ENCAPSULATE NIL
3660 implementation, at least on X86 systems; and I just noticed that
3661 encapsulation also seems closer to the spirit of the ANSI
3664 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3666 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3667 up properly again. (There was a packaging bug in 0.7.0 which
3668 left their definitions in SB-SYS::LOAD-FOREIGN and
3669 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3670 vital for most things which interface to C-level interfaces,
3671 like extensions working with sockets or databases or
3672 Perl-compatible regexes or whatever, and the need to fix
3673 this bug is the main reason that 0.7.1 was released so
3675 ** DEFGENERIC is now choosier about the methods it redefines, so that
3676 reLOADing a previously-LOADed file containing DEFGENERICs does
3677 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3678 cycle now works as it should. (thanks to Alexey Dejneka)
3679 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3680 fixed by Christophe Rhodes. (He actually submitted this patch
3681 months ago, and I delayed until after 0.7.0.)
3682 ** Bug 111 (internal compiler confusion about runtime checks on
3683 FUNCTION types) was fixed by Alexey Dejneka.
3684 * Some internal cleanups (getting rid of variables which aren't
3685 needed now that the byte interpreter is gone) caused the fasl
3686 file format number to change again.
3688 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3689 * major incompatible change: The default fasl file extension, i.e. the
3690 default extension for files produced by COMPILE-FILE, has changed
3691 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3693 ** There are many changes in the implementation of the compiler.
3694 SBCL is now essentially a compiler-only implementation of ANSI
3695 Common Lisp. EVAL still "interprets" a few special cases, but
3696 almost all the interesting cases are handled by creating
3697 a LAMBDA expression, calling COMPILE on it, then calling
3698 FUNCALL on the result.
3699 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3700 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3701 Since the code is newer, there might still be some new bugs
3702 (though not as many as before Martin Atzmueller's fixes:-). But
3703 the new code is substantially simpler and clearer, and hopefully
3704 any remaining bugs will be simpler, less fundamental, and more
3705 fixable then the bugs in the old code.
3706 ** The revised compiler is still a little unsteady on its feet.
3708 *** The debugging information it produces (particularly the names
3709 of FUNCTION objects) is sometimes much less useful than what
3710 the old compiler produced.
3711 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3712 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3713 within a MACROLET) has been temporarily weakened.
3714 ** There are new compiler optimizations for various functions:
3715 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3716 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3717 *** the math functions TRUNCATE, FLOOR, and CEILING
3718 *** the function-of-all-trades COERCE
3719 Mostly these should be transparent, but there's one
3720 potentially-annoying problem (bug 117): when the compiler
3721 inline-expands a function and does type analysis on the result,
3722 it can create control paths which have type mismatches, and
3723 when it can't prove that those control paths aren't taken,
3724 it will issue WARNINGs about the type mismatches. This is
3725 a particular problem in practice for the new sequence functions.
3726 It's not clear how this should be fixed, and for now, a
3727 workaround is given in the entry for 117 in the BUGS file.
3728 ** (Because of the interaction between the two previous items --
3729 occasional inlining problems and new inline expansions -- some
3730 of the new sequence function optimizations won't really kick in
3731 completely until debugging information, and then inlining, are
3732 straightened out in some future version.)
3733 * minor incompatible changes:
3734 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3735 DIRECTORY no longer implicitly promotes NIL slots of its
3736 pathname argument to :WILD. In particular, when you ask for the
3737 contents of a directory (which you used to be able to do without
3738 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3739 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3740 ** changes in behavior that ANSI explicitly defines to be
3741 implementation dependent:
3742 *** The new compiler-only implementation still conforms with ANSI,
3743 but acts a little different than before. Besides the obvious
3744 changes in performance tradeoffs (that the cost per form passed
3745 to EVAL has gone up, and the cost per form executed by EVAL
3746 has gone down), the behavior of the system changes a little
3747 because there are no longer any interpreted function objects.
3748 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3749 e.g. doing COMPILE on the output of interactive DEFUN is
3751 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3753 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3754 has changed from (:CL) to NIL.
3755 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3756 e.g. #\Newline and #\Tab, don't have names specified in the
3757 ANSI Common Lisp standard, is now based on their ASCII symbolic
3758 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3759 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3760 are no longer used for output.
3761 ** changes in internal implementation constants:
3762 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3763 4 million. (If your application spends a lot of time GCing and
3764 you have a lot of RAM, you might want to experiment with
3765 increasing it even more.)
3766 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3767 However, almost all old code should still continue to work without
3768 immediate update, as SB-C-CALL is now a (deprecated) nickname
3770 ** Old operator names in the style DEF-FOO are now deprecated in
3771 favor of new corresponding names DEFINE-FOO, for consistency with
3772 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3773 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3774 internal symbols, but a few supported extensions like
3775 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3776 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3777 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3778 etc. as you get deeper into recursive calls to the debugger
3779 command loop, instead of the old "5]", "5]]", "5]]]"
3780 sequence. (I was motivated to do this when squabbles between
3781 ILISP and SBCL left me very deeply nested in the debugger. In the
3782 short term, this change will probably provoke more ILISP/SBCL
3783 squabbles, but hopefully it will be an improvement in the long run.)
3784 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3785 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3786 function, since declaiming return types involving aliens is
3787 (1) annoyingly messy to do by hand and (2) vital to efficient
3788 compilation of code which calls such functions.
3789 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3790 longer reexported by the SB-EXT package. They're solely useful
3791 for alien code, so it seems more logical that you should get
3792 them from the SB-ALIEN package, not in SB-EXT.
3793 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3794 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3795 features. Instead, the code that they used to control is always
3796 built into the system.
3797 * many other bug fixes
3798 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3799 advantage of the new EVAL-WHEN stuff and to clean them up in
3800 general, and they are now more ANSI-compliant in a number of
3801 ways. Martin Atzmueller is responsible for a lot of this.
3802 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3804 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3805 *** correct ERROR type for various file operations
3806 *** some fixes for Lisp streams
3807 *** DEFMETHOD syntax checking
3808 *** changing old weird representation of debug information as
3809 strings (which, among their other deficiencies, don't transform
3810 correctly when you rename packages, and don't change their
3811 print representation when you change things like *PACKAGE*
3812 and *PRINT-LENGTH*) to symbols and lists of symbols
3813 He also made several improvements and fixed several bugs in DESCRIBE.
3814 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3816 *** misbehavior of WRITE-STRING/WRITE-LINE
3817 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3818 and several other LOOP problems as well
3819 *** DIRECTORY when similar filenames are present
3820 *** DEFGENERIC with :METHOD options
3821 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3822 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3823 *** argument ordering in FIND with :TEST option
3824 *** mishandled package designator argument in APROPOS-LIST
3825 *** various problems in the backquote readmacro
3826 *** a bug in APROPOS
3827 *** probably some others that I'm not describing very well here,
3828 since the CVS log documents them by reference to sbcl-devel
3829 messages, and the SourceForge archives aren't working well.:-(
3830 ** Dan Barlow improved the Alpha port (and is making progress on the
3831 PPC port, for those of you who think different).
3832 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3833 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3834 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3835 continues to work on the SPARC port (for those of you in a position
3836 to look down upon our little PC-compatible boxes from a great height).
3837 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3838 and uses that copy, rather than the current dynamic value, when
3839 it's trying to decide whether to truncate output. Thus e.g.
3840 (let ((*print-lines* 50))
3841 (pprint-logical-block (stream nil)
3843 (let ((*print-lines* 8))
3844 (print (aref possiblybigthings i) stream)))))
3845 should now truncate the logical block only at 50 lines, instead of
3846 often truncating it at 8 lines, as it did before.
3847 * The doc/cmucl/ directory, containing old CMU CL documentation
3848 from the time of the fork, is no longer part of the base system.
3849 SourceForge has shut down its anonymous FTP service, and with it
3850 my original plan for distributing the old CMU CL documentation
3851 there. For now, if you need these files you can download an old
3852 SBCL source release and extract them from it.
3853 * The fasl file version number changed again, for dozens of reasons,
3854 some of which are apparent above.
3856 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3857 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3858 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3859 patch, so that DISASSEMBLE output is much nicer.
3860 * The code in the SB-PROFILE package now seems reasonably stable.
3861 I still haven't decided what the final interface should look like
3862 (I'd like PROFILE to interact cleanly with TRACE, since both
3863 facilities use function encapsulation) but if you have a need
3864 for profiling now, you can probably use it successfully with
3865 the current CMU-CL-style interface.
3866 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3867 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3868 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3869 link now returns the dangling link itself, and for similar
3870 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3871 link itself. (In these cases the old code signalled an error and
3872 looped endlessly, respectively.) Thus, DIRECTORY now works even
3873 in the presence of dangling and cyclic symbolic links.
3874 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3875 is now a supported extension again, since the consensus on
3876 sbcl-devel was that it can be useful for ordinary development
3877 work, not just for debugging SBCL itself.
3878 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3879 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3880 that currently-defined functions might be redefined later with
3881 different return types.
3882 * Hash tables can be printed readably, as inspired by CMU CL code
3883 of Eric Marsden and SBCL code of Martin Atzmueller.
3884 * better error handling in CLOS method combination, thanks to
3885 Martin Atzmueller porting Pierre Mai's CMU CL patches
3886 * more overflow fixes for >16Mbyte I/O buffers
3887 * A bug in READ has been fixed, so that now a single Ctrl-D
3888 character suffices to cause end-of-file on character streams.
3889 In particular, now you only need one Ctrl-D at the command
3890 line (not two) to exit SBCL.
3891 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3893 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3894 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3895 does the right thing.
3896 * The implementation of some type tests, especially for CONDITION
3897 types, is now tidier and maybe faster, due to CMU CL code
3898 originally by Douglas Crosher, ported by Martin Atzmueller.
3899 * Some math functions have been fixed, and there are new
3900 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3901 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3902 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3903 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3904 likely to be built into sbcl-0.7.0.)
3905 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3906 CL, ported to SBCL by Martin Atzmueller)
3907 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3908 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3909 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3910 * There's a new slam.sh hack to shorten the edit/compile/debug
3911 cycle for low-level changes to SBCL itself, and a new
3912 :SB-AFTER-XC-CORE target feature to control the generation of
3913 the after-xc.core file needed by slam.sh.
3914 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3915 COMPILE-FILE is no longer supported, so that now every function
3916 gets an entry point, so that block compilation looks a little
3917 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3918 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3919 returns the number of bytes consed since the system started,
3920 rather than the number consed since the first time the function
3921 was called. (The new definition parallels ANSI functions like
3922 CL:GET-INTERNAL-RUN-TIME.)
3923 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3924 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3925 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3926 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3927 * Fasl file version numbers are now independent of the target CPU,
3928 since historically most system changes which required version
3929 number changes have affected all CPUs equally. Similarly,
3930 the byte fasl file version is now equal to the ordinary
3933 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3934 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3935 is no longer recognized. I apologize for this, because it was
3936 listed in SB-EXT as a supported extension, but I found that
3937 its existing behavior was poorly specified, as well as incorrectly
3938 specified, and it looked like too much of a mess to straighten it
3939 out. I have enough on my hands trying to get ANSI stuff to work..
3940 * many patches ported from CMU CL by Martin Atzmueller, with
3941 half a dozen bug fixes in pretty-printing and the debugger, and
3942 half a dozen others elsewhere
3943 * fixed bug 13: Floating point infinities are now supported again.
3944 They might still be a little bit flaky, but thanks to bug reports
3945 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3946 as flaky as they were.
3947 * The --noprogrammer command line option is now supported. (Its
3948 behavior is slightly different in detail from what the old man
3949 page claimed it would do, but it's still appropriate under the
3950 same circumstances that the man page talks about.)
3951 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3952 are now supported, and enabled by default. Thus, the compiler can
3953 handle many floating point and complex operations much less
3954 inefficiently. (Thus e.g. you can implement a complex FFT
3956 * The compiler now detects type mismatches between DECLAIM FTYPE
3957 and DEFUN better, and implements CHECK-TYPE more correctly, and
3958 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3959 thanks to patches from Martin Atzmueller.
3960 * various fixes to make the cross-compiler more portable to
3961 ANSI-conforming-but-different cross-compilation hosts (notably
3962 Lispworks for Windows, following bug reports from Arthur Lemmens)
3963 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3964 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3965 fixed thanks to Pierre Mai's CMU CL patches.
3966 * fixes to keep the system from overflowing internal counters when
3967 it tries to use i/o buffers larger than 16M bytes
3968 * fixed bug 45a: Various internal functions required to support
3969 complex special functions have been merged from CMU CL sources.
3970 (When I was first setting up SBCL, I misunderstood a compile-time
3971 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3972 * improved support for type intersection and union, fixing bug 12
3973 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3974 more obscure bugs as well
3975 * some steps toward byte-compiling non-performance-critical
3976 parts of the system, courtesy of patches from Martin Atzmueller
3977 * Christophe Rhodes has made some debian packages of sbcl at
3978 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3979 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3980 stable, but are nonetheless usable. When he's ready, I'd be happy
3981 to add them to the SourceForge "File Releases" section. (And if
3982 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3983 * new fasl file format version number (because of changes in
3984 internal representation of (OR ..) types to accommodate the new
3985 support for (AND ..) types, among other things)
3987 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3988 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3990 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3992 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3993 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3994 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3995 * DESCRIBE now works on structure objects again.
3996 * Most function call argument type mismatches are now handled as
3997 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3998 know whether the function will be redefined before the call is
3999 executed. (The compiler could flag local calls with full WARNINGs,
4000 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4001 it doesn't keep track of enough information to know whether calls
4002 are local in this sense.)
4003 * Compiler output is now more verbose, with messages truncated
4004 later than before. (There should be some supported way for users
4005 to override the default verbosity, but I haven't decided how to
4006 provide it yet, so this behavior is still controlled by the internal
4007 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4008 src/compiler/ir1util.lisp.)
4009 * Fasl file format version numbers have increased again, because
4010 support for the Gray streams extension changes the layout of the
4011 system's STREAM objects.
4012 * The Gray subclassable streams extension now works, thanks to a
4013 patch from Martin Atzmueller.
4014 * The full LOAD-FOREIGN extension (not just the primitive
4015 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4016 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4017 but like most other programs, it defaults to copying the Unix
4018 environment from the original process instead of starting the
4019 new process in an empty environment.
4020 * Extensions which manipulate the Unix environment now support
4021 an :ENVIRONMENT keyword option which doesn't smash case or
4022 do other bad things. The CMU-CL-style :ENV option is retained
4023 for porting convenience.
4024 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4025 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4027 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4029 * A patch from Martin Atzmueller seems to have solved the SIGINT
4030 problem, and as far as we know, signal-handling now works cleanly.
4031 (If you find any new bugs, please report them!)
4032 * The system no longer defaults Lisp source file names to types
4033 ".l", ".cl", or ".lsp", but only to ".lisp".
4034 * The compiler no longer uses special default file extensions for
4035 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4036 seems to expect a single default extension for all compiled code,
4037 and there's no compelling reason to try to stretch the standard
4038 to allow two different extensions.) Instead, byte-compiled files
4039 default to the same extension as native-compiled files.
4040 * Fasl file format version numbers have increased again, because
4041 a rearrangement of internal implementation packages made some
4042 dumped symbols in old fasl files unreadable in new cores.
4043 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4044 many fewer weird special cases.
4045 * Bug #17 (differing COMPILE-FILE behavior between logical and
4046 physical pathnames) has been fixed, and some related misbehavior too,
4047 thanks to a patch from Martin Atzmueller.
4048 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4049 by Tim Moore, ported to SBCL by Martin Atzmueller.
4050 * Martin Atzmueller fixed several filesystem-related problems,
4051 including bug #36, in part by porting CMU CL patches, which were
4052 written in part by Paul Werkowski.
4053 * More compiler warnings in src/runtime/ are gone, thanks to
4054 more patches from Martin Atzmueller.
4055 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4058 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4060 * DESCRIBE now works on CONDITION objects.
4061 * The debugger now handles errors which arise when trying to print
4062 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4064 * The build system now uses an additional file, customize-target-features.lisp,
4065 to allow local modifications to the target *FEATURES* list. (The point of
4066 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4067 debugging features enabled, without having to worry about propagating it
4068 into everyone's system when I do a "cvs update".) When no
4069 customize-target-features.lisp file exists, the target *FEATURES* list
4070 should be constructed the same way as before.
4071 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4072 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4073 warn about behavior which is undefined under the ANSI spec. Note: This
4074 is specified by ANSI, but it's not very popular with programmers.
4075 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4076 macro in the SBCL sources for an example of a workaround which you
4077 might use to make portable ANSI-standard code which does what you want.
4078 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4079 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4080 when it's used as a non-toplevel form. (This is required in order
4081 to implement the DEFCONSTANT-EQX macro.)
4082 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4083 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4084 funny things, due to bugs in the implementation of EVAL-WHEN
4085 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4086 non-toplevel DEFCONSTANTs are uncommon.)
4087 * The core file version number and fasl file version number have been
4088 incremented, because the old noncompliant DEFCONSTANT behavior involved
4089 calling functions which no longer exist, and because I also took the
4090 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4091 * fixed bug 1 (error handling before read-eval-print loop starts), and
4092 redid debugger restarts and related debugger commands somewhat while
4094 ** The QUIT debugger command is gone, since it did something
4095 rather different than the SB-EXT:QUIT command, and since it never
4096 worked properly outside the main toplevel read/eval/print loop.
4097 Invoking the new TOPLEVEL restart provides the same functionality.
4098 ** The GO debugger command is also gone, since you can just invoke
4099 the CONTINUE restart directly instead.
4100 ** The TOP debugger command is also gone, since it's redundant with the
4101 FRAME 0 command, and since it interfered with abbreviations for the
4103 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4105 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4106 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4107 some porting work by Martin Atzmueller)
4108 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4109 with a boatload of patches which he ported to SBCL. Now that those
4111 ** The system tries to make sure that its low-priority messages
4112 are prefixed by semicolons, to help people who like to use
4113 syntax highlighting in their ILISP buffer. (This patch
4114 was originally due to Raymond Toy.)
4115 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4116 patches originally written by Raymond Toy.
4117 ** The compiler understands coercion between single-value and
4118 multiple-VALUES type expressions better, getting rid of some very
4119 weird behavior, thanks to patches originally by Robert MacLachlan
4120 and Douglas Crosher.
4121 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4122 lambda lists, thanks to a patch originally by Pierre Mai.
4123 ** The system no longer bogusly warns about "abbreviated type
4125 ** The compiler gets less confused by inlining and RETURN-FROM,
4126 thanks to some patches originally by Tim Moore.
4127 ** The system no longer hangs when dumping circular lists to fasl
4128 files, thanks to a patch originally from Douglas Crosher.
4129 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4130 undefined function error.
4131 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4132 in structure slot initforms) for now, documented workaround instead:-|
4133 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4134 * fixed bug 5: added stubs for various Gray stream functions called
4135 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4136 installed, at least appropriate type errors are generated
4137 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4138 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4139 consistently in DEFMETHOD forms.
4140 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4141 it doesn't seem to affect SBCL after all
4142 * The C runtime system now builds with better optimization and many
4143 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4145 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4147 * The system is now under CVS at SourceForge (instead of the
4148 CVS repository on my home machine).
4149 * The new signal handling code has been tweaked to treat register
4150 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4151 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4152 downstream, e.g. in debug-int.lisp, has implicit dependencies
4153 on the unsignedness of integer representation of machine words,
4154 and that caused the system to bomb out with infinite regress
4155 when trying to recover from type errors involving signed values,
4156 e.g. (BUTLAST '(1 2 3) -1).)
4157 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4158 (This was one of the bugs Peter Van Eynde reported back in July.)
4159 * The system now uses code inspired by Colin Walters' O(N)
4160 implementation of MAP (from the cmucl-imp@cons.org mailing
4161 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4162 inline the MAP operation, and there is more than one
4163 sequence argument to the MAP call (so that it can't just
4164 do ETYPECASE once and for all based on the type of the
4165 single sequence argument). (The old non-inline implementation
4166 of the general M-argument sequence-of-length-N case required
4167 O(M*N*N) time when any of the sequence arguments were LISTs.)
4168 * The QUIT :UNIX-CODE keyword argument has been renamed to
4169 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4170 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4172 * Raymond Wiker's patches to port dynamic loading from Linux to
4173 FreeBSD have been added.
4174 * The BUGS file is now more nearly up to date, thanks in large part
4175 to Martin Atzmueller's review of it.
4176 * The debugger now flushes standard output streams before it begins
4177 its output ("debugger invoked" and so forth).
4178 * The core version number and fasl file version number have both
4179 been incremented, because of incompatible changes in the layout
4181 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4182 used to not be called for a saved Lisp image.)
4183 * Martin Atzmueller's version of a patch to fix a compiler crash,
4184 as posted on sbcl-devel 13 September 2000, has been installed.
4185 * Instead of installing Martin Atzmueller's patch for the
4186 compiler transform for SUBSEQ, I deleted the compiler transform,
4187 and transforms for some similar consing operations.
4188 * A bug in signal handling which kept TRACE from working on OpenBSD
4190 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4191 away by constant folding
4192 * The system now defines its address space constants in one place
4193 (in the Lisp sources), and propagates them automatically elsewhere
4194 (through GENESIS and the sbcl.h file). Therefore, patching the
4195 address map is less unnecessarily tedious and error-prone. The
4196 Lisp names of address space constants have also been systematized.
4197 * CVS tags like dollar-Header-dollar have been removed from
4198 the sources, because they have never saved me trouble and
4199 they've been source of trouble working with patches and other
4200 diff-related operations.
4201 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4202 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4204 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4206 * The system has been ported to OpenBSD.
4207 * The system now compiles with a simple "sh make.sh" on the systems
4208 that it's supported on. I.e., now you no longer need to tweak
4209 text in the target-features.lisp-expr and symlinks in src/runtime/
4210 by hand, the make.sh takes care of it for you.
4211 * The system is no longer so grossly inefficient when compiling code
4212 involving vectors implemented as general (not simple) vectors (VECTOR T),
4213 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4214 which dares to use the various sequence functions on non-simple
4215 vectors, takes less of a performance hit.
4216 * There is now a primitive type predicate VECTOR-T-P
4217 to test for the (VECTOR T) type, so that e.g.
4218 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4219 can now be compiled with some semblance of efficiency. (The old code
4220 turned the type declaration into a full call to %TYPEP at runtime!)
4221 * AREF on (VECTOR T) is still not fast, since it's still compiled
4222 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4223 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4224 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4225 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4226 without making full calls to SUBTYPEP at runtime.
4227 (Some analogous efficiency issues for non-simple vectors specialized to
4228 element types other than T, or for non-simple multidimensional arrays,
4229 have not been addressed. They could almost certainly be handled the
4230 same way if anyone is motivated to do so.)
4231 * The changes in array handling break binary compatibility, so
4232 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4233 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4234 instead of (VALUES T T).
4235 * By following the instructions that Dan Barlow posted to sbcl-devel
4236 on 2 July 2000, I was able to enable primitive dynamic object
4237 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4238 functionality is not implemented (since it calls ld to resolve
4239 library references automatically, requiring RUN-PROGRAM for its
4240 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4241 try to resolve library references) is now supported.
4242 * The system now flushes the standard output streams when it terminates,
4243 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4244 them at several other probably-convenient times, e.g. in each pass of
4245 the toplevel read-eval-print loop, and after evaluating a form given
4246 as an "--eval" command-line option. (These changes were motivated by a
4247 discussion of stream flushing issues on cmucl-imp in August 2000.)
4248 * The source transform for TYPEP of array types no longer assumes
4249 that an array whose element type is a not-yet-defined type
4250 is implemented as an array of T, but instead punts, so that the
4251 type will be interpreted at runtime.
4252 * There is now some support for cross-compiling in make.sh: each of
4253 the phases of make.sh has its own script. (This should be transparent
4254 to people doing ordinary, non-cross-compile builds.)
4255 * Since my laptop doesn't have hundreds of megabytes of memory like
4256 my desktop machine, I became more motivated to do some items on
4257 my to-do list in order to reduce the size of the system a little:
4258 ** Arrange for various needed-only-at-cold-init things to be
4259 uninterned after cold init. To support this, those things have
4260 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4261 symbols with such names are now uninterned after cold init).
4262 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4263 fasl files for cold load.
4264 ** Remove the old compiler structure pooling code (which used to
4265 be conditional on the target feature :SB-ALLOC) completely.
4266 ** Redo the representation of some data in cold init to be more compact.
4267 (I also looked into supporting byte compiled code at bootstrap time,
4268 which would probably reduce the size of the system a lot, but that
4269 looked too complicated, so I punted for now.)
4270 * The maximum signal nesting depth in the src/runtime/ support code has
4271 been reduced from 4096 to 256. (I don't know any reason for the very
4272 large old value. If the new smaller value turns out to break something,
4273 I'll probably just bump it back up.)
4274 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4276 * Many, many bugs reported by Peter Van Eynde have been added to
4277 the BUGS list; some have even been fixed.
4278 * While enabling dynamic object file loading, I tried to make the
4279 code easier to understand, renaming various functions and variables
4280 with less ambiguous names, and changing some function calling
4281 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4282 * While trying to figure out how to do the OpenBSD port, I tried to
4283 clean up some of the code in src/runtime/. In particular, I dropped
4284 support for non-POSIX signal handling, added various comments,
4285 tweaked the code to reduce the number of compilation warnings, and
4286 renamed some files to increase consistency.
4287 * To support the new automatic configuration functionality in make.sh,
4288 the source file target-features.lisp-expr has been replaced with the
4289 source file base-target-features.lisp-expr and the machine-generated
4290 file local-target-features.lisp-expr.
4291 * fixed a stupid quoting error in make.sh so that using CMU CL
4292 "lisp -batch" as cross-compilation host works again
4294 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4296 * DESCRIBE no longer tries to call itself recursively to describe
4297 bound/fbound values, so that it no longer fails on symbols which are
4298 bound to themselves (like keywords, T, and NIL).
4299 * DESCRIBE now works on generic functions.
4300 * The printer now prints less-screwed-up representations of closures
4301 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4302 * A private symbol is used instead of the :EMPTY keyword previously
4303 used to mark empty slots in hash tables. Thus
4304 (DEFVAR *HT* (MAKE-HASH-TABLE))
4305 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4306 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4307 now does what ANSI says that it should. (You can still get
4308 similar noncompliant behavior if bang on the hash table
4309 implementation with all the symbols you get back from
4310 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4311 This breaks binary compatibility, since tests for equality to
4312 :EMPTY are wired into things like the macroexpansion of
4313 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4315 * There's now a minimal placeholder implementation for CL:STEP,
4316 as required by ANSI.
4317 * An obscure bug in the interaction of the normal compiler, the byte
4318 compiler, inlining, and structure predicates has been patched
4319 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4320 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4322 * Missing ordinary arguments in a macro call are now detected even
4323 when the macro lambda list contains &KEY or &REST.
4324 * The debugger no longer complains about encountering the top of the
4325 stack when you type "FRAME 0" to explicitly instruct it to go to
4326 the top of the stack. And it now prints the frame you request even
4327 if it's the current frame (instead of saying "You are here.").
4328 * As specified by ANSI, the system now always prints keywords
4329 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4331 * The default initial SIZE of HASH-TABLEs is now smaller.
4332 * Type information from CLOS class dispatch is now propagated
4333 into DEFMETHOD bodies, so that e.g.
4334 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4336 is now basically equivalent to
4337 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4338 (DECLARE (TYPE SINGLE-FLOAT X))
4340 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4341 operation, without having to do run-time type dispatch.
4342 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4343 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4344 * Since I don't seem to be making big file reorganizations very often
4345 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4346 is overflowing my ability to conveniently back them up), I've finally
4347 checked the system into CVS. (The CVS repository is on my home system,
4348 not at SourceForge -- putting it on SourceForge might come later.)
4349 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4350 high-level GC-NOTIFY-FOO functions send their output. (There's
4351 still very little control of where low-level verbose GC functions
4352 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4353 less than it used to -- the GC-NOTIFY-FOO functions are now under
4354 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4355 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4356 in only one place in the source code, and propagates it automatically
4357 everywhere that it's needed. Thus e.g. when I bump the version from
4358 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4359 * The C source files now include boilerplate legalese and documentation
4360 at the head of each file (just as the Lisp source files already did).
4361 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4362 to his page will be replaced with a link to his new CLiki service.
4364 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4366 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4367 * The build process now looks for GNU make under the default name "gmake",
4368 instead of "make" as it used to. If GNU make is not available as "gmake"
4369 on your system, you can change this default behavior by setting the
4370 GNUMAKE environment variable.
4371 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4372 can build without error under CMU CL.
4374 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4376 * There is now a partial SBCL user manual (with some new text and some
4377 text cribbed from the CMU CL manual).
4378 * The beginnings of a profiler have been added (starting with the
4379 CMU CL profiler and simplifying and cleaning up). Eventually the
4380 main interface should be through the TRACE macro, but for now,
4381 it's still accessed through vaguely CMU-CL-style functions and macros
4382 exported from the package SB-PROFILE.
4383 * Some problems left over from porting CMU CL to the new
4384 cross-compilation bootstrap process have been cleaned up:
4385 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4386 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4387 part when getting rid of the compiler *BACKEND* stuff.)
4388 ** The value of *NULL-TYPE* was screwed up, because it was
4389 being initialized before the type system knew the final
4390 definition of the 'NULL type. This screwed up several key
4391 optimizations in the compiler, causing inefficiency in all sorts
4392 of places. (I found it because I wanted to understand why
4393 GET-INTERNAL-RUN-TIME was consing.)
4394 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4395 PROCLAIM FTYPE information. Unfortunately this broke binary
4396 compatibility again, since now the forms output by DEFGENERIC
4397 to refer to functions which didn't exist in 0.6.3.
4398 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4399 can use the new (as of 0.6.3) transform for SOME into MAP into
4401 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4402 values to (INTEGER 0 1000000), so that the time code will no longer
4403 occasionally get blown up by Linux returning 1000000 microseconds
4404 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4405 its output conform to the ANSI spec. (Alas, this makes its output
4406 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4407 * A full call to MAP NIL with a single sequence argument no longer conses.
4408 * fixes to problems pointed out by Martin Atzmueller:
4409 * The manual page no longer talks about multiprocessing as though
4410 it were currently supported.
4411 * The ILISP support patches have been removed from the distribution,
4412 because as of version 5.10.1, ILISP now supports SBCL without us
4413 having to maintain patches.
4414 * added a modified version of Raymond Toy's recent CMU CL patch for
4415 EQUALP comparison of HASH-TABLE
4417 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4419 * The system still can't cross-compile itself with
4420 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4421 entails), but at least it can compile more of itself that way
4422 than it used to be able to, and various buglets which were uncovered
4423 by trying to cross-compile itself that way have now been fixed.
4424 * This release breaks binary compatibility again. This time
4425 at least I've incremented the FASL file format version to 2, so that the
4426 problem can be detected reliably instead of just causing weird errors.
4427 * various new style warnings:
4428 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4429 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4430 ** using the lexical binding of a variable named in the *FOO* style
4431 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4432 as specified by ANSI.
4433 * *RANDOM-STATE* is no longer automatically initialized from
4434 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4435 default behavior of the system is to repeat its behavior every time
4436 it's run. If you'd like to change this behavior, you can always
4437 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4438 old convention there was no comparably easy way to get the system to
4439 repeat its behavior every time it was run.
4440 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4441 FTYPE declarations has been removed, in favor of their ANSI
4442 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4443 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4444 the types of their sequence arguments can be determined at compile time.
4445 This is done through a new open code expansion for MAP which eliminates
4446 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4447 argument types can be determined at compile time.
4448 * The optimizer now transforms COERCE into an identity operation when it
4449 can prove that the coerced object is already of the correct type. (This
4450 can be a win for machine generated code, including the output of other
4451 optimization transforms, such as the MAP transform above.)
4452 * Credit information has been moved from source file headers into CREDITS.
4453 * Source file headers have been made more standard.
4454 * The CASE macro now compiles without complaining even when it has
4457 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4459 * (Note that the way that the PCL macroexpansions were rewritten
4460 to accommodate the change in DEFGENERIC below breaks binary
4461 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4462 not run under sbcl-0.6.2. Once we get out of alpha releases,
4463 i.e. hit release 1.0.0, we'll probably try to maintain binary
4464 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4465 and sbcl-1.4.4. Until then, however, it might be fairly common
4466 for maintenance releases to break binary compatibility.)
4467 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4469 * The reporting of unhandled signals has been changed to print some
4470 explanatory text as well as the report form. (Previously only
4471 the report form was printed.)
4472 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4473 it defines, so that the compiler no longer issues undefined function
4474 warnings for compiled-but-not-yet-loaded generic functions.
4475 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4476 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4477 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4478 section of the ANSI spec are supported.
4479 * The "" nickname for the "KEYWORD" package has been removed.
4480 The reader still handles symbol tokens which begin with a package marker
4481 as keywords, but it doesn't expose its mechanism for doing so in the
4482 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4483 * The system now issues STYLE-WARNINGs for contradictory TYPE
4484 proclamations. (Warnings for contradictory FTYPE proclamations would
4485 be nice too, but those can't be done usefully unless the type system
4486 is made smarter about FUNCTION types.)
4487 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4488 systematized, so that "*target-*.lisp is supposed to exist only on the
4489 target and imply that there's a related file which exists on the
4490 host, and *host-*.lisp is supposed to exist only on the host and imply
4491 that there's a related file which exists on the target. This involves a
4492 lot of renaming. Hopefully the acute confusion caused by the renaming
4493 will be justified by the reduction in chronic confusion..
4494 ** runtime-type.lisp -> early-target-type.lisp
4495 ** target-type.lisp -> late-target-type.lisp
4496 ** early-host-format.lisp -> early-format.lisp
4497 ** late-host-format.lisp -> late-format.lisp
4498 ** host-error.lisp -> misc-error.lisp
4499 ** early-error.lisp -> early-target-error.lisp
4500 ** late-error.lisp -> late-target-error.lisp
4501 ** host-defboot.lisp -> early-defboot.lisp
4502 ** code/misc.lisp -> code/target-misc.lisp
4503 ** code/host-misc.lisp -> code/misc.lisp
4504 ** code/numbers.lisp -> code/target-numbers.lisp
4505 ** code/early-numbers.lisp -> numbers.lisp
4506 ** early-host-type.lisp -> early-type.lisp
4507 ** late-host-type.lisp -> late-type.lisp
4508 ** host-typep.lisp -> typep.lisp
4509 ** load.lisp -> target-load.lisp
4510 ** host-load.lisp -> load.lisp
4511 ** host-disassem.lisp -> disassem.lisp
4512 ** host-insts.lisp -> insts.lisp
4513 ** byte-comp.lisp -> target-byte-comp.lisp
4514 ** host-byte-comp.lisp -> byte-comp.lisp
4515 ** host-signal.lisp -> signal.lisp
4516 ** host-defstruct.lisp -> defstruct.lisp
4517 ** late-target-type.lisp -> deftypes-for-target.lisp
4518 Furthermore, several other previously target-only files foo.lisp (e.g.
4519 hash-table.lisp and random.lisp) have been split into a target-and-host
4520 foo.lisp file and a target-only target-foo.lisp file, with their key type
4521 definitions in the target-and-host part, so that the cross-compiler will
4522 know more about target types.
4523 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4524 gone away. In their place are various *BACKEND-FOO* variables
4525 corresponding to the slots of the old structure.
4526 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4527 into the target SBCL has been fixed.
4528 * The chill.lisp system for loading cold code into a running SBCL
4530 * Support for the CMU CL "scavenger hook" extension has been removed.
4531 (It was undocumented and unused in the CMU CL sources that SBCL was
4532 derived from, and stale in sbcl-0.6.1.)
4533 * Various errors in the cross-compiler type system were detected
4534 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4535 (enabling various consistency checks). Many of them were fixed,
4536 but some hard problems remain, so the compiler is back to
4537 running without *TYPE-SYSTEM-INITIALIZED* for now.
4538 * As part of the cross-compiler type system cleanup, I implemented
4539 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4540 * I have started adding UNCROSS calls throughout the type system
4541 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4542 on cross-compiler input files will be able to go away, and various
4544 * CONSTANTP now returns true for quoted forms (as explicitly required
4547 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4549 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4550 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4551 (MAKE-LIST -1). In the longer term, it ought to become true
4552 that declarations are assertions even at SAFETY 1. For now, it's not
4553 quite true even at SAFETY 3, but it's at least more nearly true..
4554 (Note that this change seems to increases the size of the system by
4555 O(5%) and to decrease the speed of the compiler by 20% or more.)
4556 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4557 for the problem of printing dozens of lines of distracting information
4558 about low-level system machinery as part of the top stack frame
4559 on entry to the debugger when an undefined function was called.
4560 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4561 is not reset to COMMON-LISP-USER.
4562 * Compilation of stuff related to dyncount.lisp has been made conditional
4563 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4564 smaller. The various dyncount-related symbols have been moved into
4565 a new "SB-DYNCOUNT" package.
4566 * tty-inspect.lisp has been renamed to inspect.lisp.
4567 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4568 feature has gone away. (When we eventually port to other flavors of
4569 libc and/or Unix, we'll try to make the differences between flavors
4570 invisible at the user level.)
4571 * Various other *FEATURES* tags, and/or their associated conditionals,
4572 have been removed if obsolescent, or given better documentation, or
4573 sometimes given more-mnemonic names.
4575 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4577 * tidied up "make.sh" script
4578 * tidied up system directory structure
4579 * better "clean.sh" behavior
4580 * added doc/FOR-CMUCL-DEVELOPERS
4581 * many many small tweaks to output format, e.g. removing possibly-confusing
4582 trailing #\. character in DESCRIBE-INSTANCE
4583 * (EQUALP #\A 'A) no longer signals an error.
4584 * new hashing code, including EQUALP hashing
4585 * tidied up Lisp initialization and toplevel
4586 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4587 * command line argument processing
4588 * added POSIX-GETENV function to deal with Unix-ish environment variables
4589 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4590 terminating SBCL on EOF
4591 * non-verbose GC by default
4592 * There is no more "sbcl" shell script; the sbcl file is now the C
4593 runtime executable (just like CMU CL).
4594 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4596 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4597 symbol and package information as Lisp native symbol and package objects
4598 instead of strings naming symbols and strings naming packages. This way,
4599 whenever packages are renamed (as in warm init), debug information is
4600 transformed along with everything else.
4601 * tweaked the optimization policy declarations which control the building
4602 of SBCL itself. Now, among other things, the system no longer saves
4603 source location debugging information. (This helps two problems at once
4604 by reducing SBCL size and by keeping SBCL from trying to look for its
4605 sources -- which may not exist -- when reporting errors.)
4606 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4607 debugging and testing purposes
4608 * cleaned up printing, making the printer call PRINT-OBJECT for
4609 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4610 methods, giving nearly-ANSI behavior
4611 * converted almost all special variables to use *FOO* naming convention
4612 * deleted PARSE-TIME functionality, since it can be done portably
4613 * moved some files out of cold init into warm init
4614 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4616 * regularized formatting of source files
4617 * added an install.sh script
4618 * fixed ridiculous memory usage of cross-compiler by making
4619 compiler/alloc.lisp not try to do pooling unless it can hook
4620 itself into the GC of the cross-compilation host. Now the system
4621 builds nicely on my old laptop.
4622 * added :SB-ALLOC in target-features.lisp-expr
4623 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4624 was not implemented)
4625 * re-did condition handling and note reporting in the compiler. Notes
4626 are no longer handled by signalling conditions. Style warnings
4627 and warnings are handled more correctly and reported in such a way
4628 that it's easy to find one or the other in your output (so that you
4629 can e.g. figure out which of many problems caused COMPILE-FILE to
4631 * changed the severity of several compiler warnings from full WARNING
4632 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4633 compiler note reporting so that it doesn't use the condition system
4634 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4635 COMPILE-FILE command)
4636 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4637 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4638 no longer supported, since their implementation was deeply intertwingled
4639 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4640 * removed bogus "support" for compiler macros named (SETF FOO), and
4641 removed the compiler macro for SETF INFO (but only after making a fool
4642 of myself on the cmucl-imp mailing list by posting a bogus patch for
4643 DEFINE-COMPILER-MACRO..)
4644 * Compiled files containing forms which have side effects on the Lisp
4645 reader (such as DEFPACKAGE forms) are now handled more correctly.
4646 (Compiler queuing of top level lambdas has been suppressed by setting
4647 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4648 * deleted various currently-unused source files, e.g. gengc.lisp. They
4649 may be added back at some point e.g. when porting to other architectures,
4650 but until they are it's distracting to distribute them and to try to
4652 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4653 there were so many of them they're just distractions, and UNCROSS is
4654 known to be able to handle the current sources
4655 * moved PROFILE functionality into TRACE, so that it will be clear
4656 how the wrapping and unwrapping of functions when you profile them
4657 interacts with the wrapping and unwrapping of functions when you
4658 trace them. (Actually, the functionality isn't there yet, but at least
4659 the interface specification is there. Hopefully, the functionality will
4660 arrive with some future maintenance release.)
4661 * removed host-oops.lisp
4662 * changed signature of QUIT function to allow UNIX-CODE argument
4663 * fixed READ-SEQUENCE bug
4664 * tweaked verbose GC output so that it looks more like the progress
4665 output that ANSI specifies for functions like LOAD
4666 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4667 * added <http://sbcl.sourceforge.com> to the banner information printed by