1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.17 relative to 1.0.16:
3 * temporary regression: user code can no longer allocate closure
4 variable storage on stack, due to bug 419 without explicitly
5 requesting it. Please consult sbcl-devel for advice if you need to
6 use this feature in the meanwhile.
7 * new feature: runtime argument --control-stack-size can be used to
8 adjust thread default control stack size.
9 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
11 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
12 * optimization: two argument forms of LAST are upto ~10% faster
13 in normal SPEED policies.
14 * optimization: NCONC no longer needs to heap cons its &REST list
15 in normal SPEED policies.
16 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
17 since 1.0.15.36. (thanks to Paul Khuong)
18 * bug fix: LAST when always returned the whole list when given a bignum
19 as the second argument.
20 * bug fix: dynamic extent allocation of nested lists and vectors
21 could leak to otherwise accessible parts.
22 * bug fix: invalid optimization of heap-allocated alien variable
24 * bug fix: fasl header checking is less vulnerable to different
25 platform word lengths.
26 * bug fix: more correct assembler syntax for GNU binutils
27 2.18.50.0.4 support. (thanks to Marijn Schouten)
28 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
31 changes in sbcl-1.0.16 relative to 1.0.15:
32 * minor incompatible change: revert the changes to sb-posix's error
33 signaling added in 1.0.14.
34 * minor incompatible change: change PROBE-FILE back to returning
35 NIL whenever we can't get a truename, as was the case before 1.0.14.
36 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
37 inherits from ERROR instead of just CONDITION.
38 * new feature: SB-INTROSPECT can provide source locations for instances
39 as well. (thanks to Tobian Ritterweiler)
40 * optimization: binding special variables now generates smaller code
41 on threaded platforms.
42 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
43 and cases where no :TEST is given but the compiler can infer that
44 the element to search is of type (OR FIXNUM (NOT NUMBER)).
45 * optimization: better LOGNOT on fixnums.
46 * optimization: modular arithmetic for a particular requested width
47 is implemented using a tagged representation unless a better
48 representation is available.
49 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
50 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
51 for wellformedness and returns a specialized vector. (reported by
53 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
54 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
55 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
56 unscheduling timer at the same time as another timer fires.
57 * bug fix: don't reschedule timers for dead threads.
58 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
59 * bug fix: copying output from RUN-PROGRAM to a stream signalled
60 bogus errors if select() was interrupted.
61 * enhancement: add support for fcntl's struct flock to SB-POSIX.
63 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
64 * enhancement: cleaner backtraces for interactive interrupts, as
65 well as other cases where the interesting frames used to be
66 obscured by interrupt handling frames.
67 * enhancement: untracing a whole package using (UNTRACE "FOO") is
68 now supported, and tracing a whole package using (TRACE "FOO") now
69 traces SETF-functions as well.
70 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
71 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
72 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
73 * unreadably printed representation of hash-tables now includes
75 * bug fix: partially fixed #188: type propagation from assignments
76 is now more efficient.
77 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
78 'DOUBLE-FLOAT) are not flushable.
79 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
80 full address of the object, and none of the tag bits.
81 * bug fix: readably printing hash-tables now respects other printer
82 control variables. (reported by Cedric St-Jean)
83 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
84 keyword with MAKE-HASH-TABLE.
85 * bug fix: export SB-POSIX:MKSTEMP.
86 * bug fix: SORT was not interrupt safe.
87 * bug fix: XREF accounts for the last node of each basic-block as
89 * bug fix: MAKE-INSTANCE optimizations interacted badly with
90 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
91 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
92 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
93 create obsolete instances in the case of redefinition or
94 obsoletion of a superclass. (thanks to Andy Hefner)
95 * bug fix: Support for the Alpha architecture has been revived; it had
96 suffered somewhat from lack of maintenance since sbcl-1.0.
97 * improvements to the Windows port:
98 ** The system detects the codepage to use at startup, instead of
99 using the value from the saved core. (thanks to Kei Suzuki)
101 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
102 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
103 (see documentation for details.)
104 * revived support for OpenBSD (contributed by Josh Elsasser)
105 * partially fixed bug #108: ROOM no longer suffers from occasional
106 (AVER (SAP= CURRENT END)) failures.
107 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
108 bodies no longer cause a WARNING to be signalled. (reported by
110 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
111 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
112 the bug in TRUNCATE.) (reported by Michael Weber)
113 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
114 no samples. (reported by Andy Hefner)
115 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
116 no longer appear as (NIL ...) frames in backtraces.
117 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
118 is now actually exported from SB-EXT as documented. (reported by
120 * bug fix: sb-aclrepl now correctly understands how to inspect
121 single-floats on 64-bit platforms where single-floats are not boxed.
122 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
123 yet been finalized. (reported by Levente Meszaros)
124 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
125 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
126 SBCL-specific optimize qualities.
128 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
129 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
130 an executable in the search path, and does so in the child
131 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
132 been removed; it can be found in the CVS history, for somebody who
133 needs that search behavior (see the manual).
134 * minor incompatible change: compiler policy re. weakening type
135 checks has changed: now type checks are weakened only if SAFETY < 2
137 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
138 unparsing of directory pathnames as files. Analogously,
139 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
140 filename to parse into a directory pathname.
141 * enhancement: implicit generic function creation now signals a
142 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
143 users can bind handlers for and muffle around calls to LOAD.
144 * enhancement: RUN-PROGRAM allows unicode arguments and environments
145 to be used (using the default stream external format), and allows
146 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
147 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
148 strings and vectors whose element-type or simplicity is not fully
149 known at compile-time.
150 * optimization: STRING-TRIM and related functions no longer allocate
151 a new string when no trimming needs to be performed. These functions
152 are also faster than before when the input string has been declared
154 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
155 * optimization: READ-LINE is significantly faster for files containing
157 * optimization: non-open coded uses of character comparison operators
158 (e.g. char=) no longer cons when called with more than one parameter
159 on platforms supporting dynamic-extent allocation.
160 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
161 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
162 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
164 * bug fix: some sequence functions elided bounds checking when
166 * bug fix: too liberal weakening of union-type checks when SPEED >
168 * bug fix: more bogus fixnum declarations in ROOM implementation
171 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
172 * new feature: MAKE-HASH-TABLE now experimentally accepts a
173 :SYNCHRONIZED argument, which makes the hash-table safe for
174 concurrent accesses (but not iteration.) See also:
175 SB-EXT:WITH-LOCKED-HASH-TABLE, and
176 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
177 * optimization: CONCATENATE on strings is an order of magnitude faster
178 in code compiled with (> SPEED SPACE).
179 * optimization: SUBSEQ is ~50% faster on lists.
180 * bug fix: bug 417 fixed -- source location reporting is now more robust.
181 * bug fix: SUBSEQ on a list will now correctly signal an error if if
182 END is smaller then START.
183 * bug fix: SB-PROFILE will no longer report extra consing for nested
184 calls to profiled functions.
185 * bug fix: ROOM implementation had bogus fixnum declarations which
186 could cause type-errors when calling ROOM on large images.
187 * bug fix: if file compilation is aborted, the partial fasl is now
188 deleted, and COMPILE-FILE returns NIL as the primary value.
189 * bug fix: number of thread safety issues relating to SBCL's internal
190 hash-table usage have been fixed.
191 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
192 be returned from its body when the values were being returned
193 using unknown-values return convection and the W-P-O was wrapped
194 inside an UNWIND-PROTECT.
195 * bug fix: sb-posix should now compile again under Windows, enabling
198 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
199 * incompatible change: hash-table accessor functions are no longer
200 automatically protected by locks. Concurrent accesses on the same hash-table
201 from multiple threads can give inconsistent results or even corrupt the
202 hash-table completely. Multi-threaded applications should do their own
203 locking at the correct granularity. In the current implementation it is
204 still safe to have multiple readers access the same table, but it's not
205 guaranteed that this property will be maintained in future releases.
206 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
207 and will signal an error at runtime.
208 * enhancement: SB-THREAD package now exports a semaphore interface.
209 * enhancement: CONS can now stack-allocate on x86 and
210 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
212 * enhancement: nested lists can now be stack allocated on
213 platforms providing stack allocation support.
214 * enhancement: dynamic-extent support has been extended to support
215 cases where there are multiple possible sources for the stack
217 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
218 if the mutex is uncontested on Linux.
219 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
220 as specified by the CLHS. (thanks to Tobias Rittweiler)
221 * bug fix: NaN comparison now works on x86-64.
222 * bug fix: CLOSE :ABORT T on a stream with pending output now
224 * bug fix: instances of non-standard metaclasses using standard
225 instance structure protocol sometimes missed the slot type checks
227 * bug fix: known functions can be inlined in high-debug code.
229 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
230 * minor incompatible change: the MSI installer on Windows no longer
231 associates .lisp and .fasl files with the installed SBCL.
232 * minor incompatible change: :UNIX is no longer present in *FEATURES*
233 on Windows. (thanks to Luis Oliviera)
234 * new feature: SB-CLTL2 contrib module now implements
235 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
236 * optimization: scavenging weak pointers is now more efficient,
237 requiring O(1) instead of O(N) per weak pointer to identify
238 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
239 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
240 method bodies using constant slot-names when the first argument is
241 a specializer parameter for the method.
242 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
243 names, when the class of the instance is a direct instance
244 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
245 as slow as the constant slot-name case.)
246 * optimization: member type construction is now O(N) instead
248 * optimization: UNION and NUNION are now O(N+M) for large
249 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
251 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
252 now more readable in environments like Slime which display it.
253 (thanks to Tobias C. Rittweiler)
254 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
255 was able to infer the correct argument type for the object on which
256 the CAS operation was being performed.
257 * bug fix: copy propagation interfered with parallel assignment
258 semantics in local calls. (reported by Paul Khuong)
259 * bug fix: the signed modular fixnum shift compiled to wrong code on
260 x86 and x86-64. (spotted by a slight modification to some of
263 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
264 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
265 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
266 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
268 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
269 and generic function now print the object identity for anonymous
270 instances. (thanks to Attila Lendvai)
271 * enhancement: as an extension to MOP, SBCL now supports SETF of
272 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
273 (thanks to Attila Lendvai)
274 * enhancement: sb-cover has an alternate annotation mode which puts
275 the form annotations at the car of the form, instead of annotating
276 it completely. The benefit of this mode is that it shows better
277 which forms were instrumented by the compiler.
278 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
279 by printer control variables. (reported by Dan Corkill)
280 * bug fix: system leaked memory when delayed output was performed by
281 the OS in smaller chunks then expected. (thanks to David Smith)
282 * bug fix: system leaked memory when file streams were not closed
284 * bug fix: large objects written to slow streams that were modified
285 after the write could end up with the modified state written to
286 the underlying file descriptor.
287 * bug fix: multiple threads operating in parallel on the same stream
288 could cause buffer-overflows.
289 * bug fix: source location information is stored correctly
290 (broken since 1.0.6). This bug would generally show up as the
291 Slime debugger highlighting the wrong form when the sldb-show-source
293 * bug fix: Branch forms are again annotated as branches in the sb-cover
295 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
296 on x86-64. (reported by Gregory Vanuxem)
297 * bug fix: EQUALP could return wrong results for structures with raw
298 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
299 word). (reported by Vjacheslav Fyodorov)
300 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
303 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
304 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
305 atomic compare-and-swap operations on threaded platforms.
306 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
307 allows assigning a global minimum value to optimization qualities
308 (overriding proclamations and declarations).
309 * enhancement: closed over variables can be stack-allocated on x86
311 * performance bug fix: GETHASH and (SETF GETHASH) are once again
313 * optimization: slot definition lookup is now O(1). This speeds up
314 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
315 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
316 * optimization: ASSOC and MEMBER can now be open-coded for all
317 combinations of keyword arguments when second argument is constant
318 and SPEED >= SPACE. In other cases a specialized version is
320 * bug fix: using obsoleted structure instances with TYPEP and
321 generic functions now signals a sensible error.
322 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
323 (reported by Kristoffer Kvello)
324 * bug fix: backtrace construction is now more careful when making
325 lisp-objects from pointers on the stack, to avoid creating bogus
326 objects that can be seen by the GC.
327 * bug fix: defaulting of values in contexts expecting more than 7
328 variables now works on x86-64. (reported by Christopher Laux)
329 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
331 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
332 as the property-list of a symbol.
333 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
334 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
335 situations requested, are once again file-compileable. (reported
338 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
339 * MOP improvement: support for user-defined subclasses of
340 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
341 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
342 the expansion of DEFMETHOD, is responsible for generating a form
343 which creates a list of specializers when evaluated. Additional
344 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
345 debugging and introspective support.
346 * minor incompatible change: the (unsupported) spinlock interface
347 has changed: free spinlock now has the value NIL, and a held spinlock
348 has the owning thread as its value.
349 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
350 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
352 * enhancement: name of a socket-stream is now "a socket" instead of
354 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
355 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
356 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
357 for generic functions. (Thanks to Tobias C. Rittweiler)
358 * optimization: bignum printing speed has been improved by 20-40%
359 (depending on the bignum size.)
360 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
362 * bug fix: the cache used by the CLOS to store precomputed effective
363 methods, slot offsets, and constant return values is now thread and
365 * bug fix: generic function dispatch function updating is now thread
366 and interrupt safe (in the sense that the known issues have been
368 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
369 * bug fix: interning EQL-specializers is now thread and interrupt safe.
370 * bug fix: asdf systems with dependencies to the SB-POSIX or
371 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
372 * bug fix: interrupt safety of applicable method computation has been
375 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
376 * new contrib: sb-cover, an experimental code coverage tool, is included
378 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
379 significantly faster.
380 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
381 produce more efficient code on x86-64 if the compiler can determine
382 that (+ INDEX <constant>) does not require a bounds check and FOO
383 has an element type at least 8 bits wide.
384 * enhancement: a new, experimental synchronous timeout facility is
385 provided. Refer to SB-SYS:WITH-DEADLINE for details.
386 * enhancement: when a symbol name conflict error arises, the
387 conflicting symbols are always printed with a package prefix.
388 (thanks to Kevin Reid)
389 * enhancement: stepping is now once again supported on the SPARC and
390 MIPS platforms. (It is also now more likely to work on CheneyGC
392 * enhancement: sb-sprof can now also track and report accurate call
394 * bug fixes: the treatment of non-standard subclasses of
395 SB-MOP:SPECIALIZER is more correct.
396 * incompatible change: PURIFY no longer copies the data from the
397 dynamic space into the static and read-only spaces on platforms
398 that use the generational garbage collector
399 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
401 * bug fix: GC race condition occasionally resulting in crashes with
402 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
404 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
406 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
407 a STYLE-WARNING to be signalled (regression from 1.0.4.)
408 * bug fix: an asynchronous interrupt could previously leave the
409 system running with GC inhibited.
410 * bug fix: a DECLARE form evaluated at top-level now causes an error
411 rather than silently (or verbosely) returning NIL.
412 * bug fix: trying to dissassemble functions compiled at high DEBUG could
413 result in a error being signalled due to source form lookup errors.
414 (reported by Peter Graves)
416 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
417 * incompatible change: removed writer methods for host-ent-name,
418 host-ent-addresses -- changing the values did not update the DNS
420 * minor incompatible change: changed experimental JOIN-THREAD interface
421 * documentation: the manual now lists reader and writer methods
422 in class slot documentation sections. (thanks to Richard M Kreuter)
423 * documentation: unwinding from asyncronous events has been
424 documented as unsafe.
425 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
426 in multithreaded application code.
427 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
428 platforms. (thanks to James Anderson for the optimization hint)
429 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
431 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
432 variants no longer cons.
433 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
434 their NOT- variants no longer cons.
435 * optimization: EQUAL hash tables no longer use SXHASH for objects
436 of all data types, but instead use an EQL hash for types for which
437 EQUAL is the same as EQL.
438 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
439 are significantly faster.
440 * optimization: new STRING-OUTPUT-STREAM implementation is much
441 faster for large outputs and conses 30% less on average.
442 * enhancement: XREF information is now collected for references made
443 to global variables using SYMBOL-VALUE and a constant argument.
444 * enhancement: SIGINT now causes a specific condition
445 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
446 * bug fix: FILE-POSITION used to signal an error in some cases where
447 ANSI requires it to return NIL.
448 * bug fix: ADJUST-ARRAY is now interrupt-safe.
449 * bug fix: adding and removing fd-handlers is now interrupt-safe.
450 * bug fix: inlined calls to C now ensure 16byte stack alignment on
452 * bug fix: bad type declaration in the CLOS implementation has
453 been fixed. (reported by James Anderson)
454 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
455 (thanks to Lutz Euler)
456 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
457 required by the CLHS. (reported by Samium Gromoff)
458 * bug fix: dead unbound variable references now signal an error.
459 (reported by Marco Monteiro)
460 * bug fix: / with an unused value was being deleted in safe code.
461 (thanks to Marco Monteiro and Kevin Reid)
462 * bug fix: number of characters that can be written onto a single
463 line in a file is unlimited.
464 * bug fix: some GC deadlocks caused by asynchronous interrupts have
465 been fixed by inhibiting interrupts for when GC is disabled.
466 * bug fix: some interrupt safety issues with GETHASH, (SETF
467 GETHASH), CLRHASH and REMHASH have been fixed.
468 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
469 specifier no longer causes infinite recursion.
470 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
471 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
472 * bug fix: an &environment argument with the correct variable information
473 is passed to macros that are expanded during byte compilation.
474 (reported by Samium Gromoff)
475 * bug fix: a logic error was causing the hash function for bignums to
476 have excessive amounts of collisions. (reported by Faré Rideau)
477 * bug fix: modifying the contents of an array could change the return
478 value of SXHASH on that array, which is only allowed for strings
479 and bit vectors. (bug introduced in 0.9.16)
480 * improvement: the x86-64/darwin port now passes all tests (except
481 for the debugger tests) but should still be considered
482 experimental until this is fixed.
483 * improvement: a style-warning is signaled for CASE (etc) clauses with
484 duplicate keys. (thanks to Kevin Reid)
485 * improvement: macos/x86 and macos/x86-64 now use mach exception
486 handlers for dealing with illegal instructions (for trapping and
487 error handling) and memory protection violations (for GC).
489 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
490 * new platform: experimental support for x86-64/darwin (MacOS).
491 * incompatible change: the thread-safe (on most platforms) getaddrinfo
492 and getnameinfo sockets functions are used instead of gethostbyaddr
493 and gethostbyname, on platforms where the newer functions are available.
494 As a result, the ALIASES field of HOST-ENT will always be NIL on these
496 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
497 don't print the startup banner, but behave as if --noinform was passed
498 as a command line argument. (thanks to Kevin Reid)
499 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
500 * optimization: code using alien values with undeclared types is much faster.
501 * optimization: the compiler is now able to open code SEARCH in more cases.
502 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
503 * bug fix: using standardized COMMON-LISP special variables as loop
504 variables no longer signals bogus package lock violations. (reported
506 * bug fix: declaring local loop variables to be of a range-limited type
507 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
508 (reported by Andras Simon)
509 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
510 bugs remain on x86-64.)
511 * bug fix: the #= and ## reader macros now interact reasonably with
512 funcallable instances.
513 * bug fix: type-checks for function arguments were compiled using the
514 compiler policy settings of the wrong lexical environment (in 1.0.2
516 * bug fix: SHADOW accepts characters as string designators, as required
517 by the spec (thanks to Eric Marsden)
518 * bug fix: fixed GC safety issues when foreign functions are called with
519 non-base strings as arguments
520 * bug fix: more consistent error output for fatal-compiler-warnings, like
522 * bug fix: functions with proper names were showing up as NIL in
523 backtraces (reported by Edi Weitz, regression in 1.0.2)
525 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
526 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
527 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
528 produce more efficient code on the x86 if the compiler can determine
529 that (+ INDEX <constant>) does not require a bounds check and FOO
530 has an element type at least 8 bits wide.
531 * bug fix: references to undefined variables are handled the same way
532 in toplevel forms as in the normal compiler
533 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
535 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
536 (thanks to Jon Buller)
537 * improvement: faster compilation times for complex functions
538 * improvement: added readlink support to SB-POSIX (thanks to Richard
541 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
542 * improvement: experimental support for mach exception handling on
543 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
545 * improvement: support for GBK external format.
546 (thanks to Chun Tian (binghe))
547 * improvement: the debugger now displays variables that have been closed
548 over, in code compiled with (DEBUG 2) or higher
549 * improvement: support for executable cores on NetBSD (thanks to
551 * new feature: added a RESTART-FRAME debugger command
552 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
553 be used to provide an implementation for FILE-POSITION on Gray streams
554 (thanks to Eric Marsden)
555 * new feature: add syslog, openlog and closelog support to SB-POSIX
556 (thanks to Richard Kreuter)
557 * optimization: the function call overhead in code compiled with
558 a high DEBUG optimization setting is significantly
559 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
560 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
561 (thanks to Tony Martinez)
562 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
563 evaluated once (reported by Kevin Reid)
564 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
565 works again (thanks to Kilian Sprotte)
566 * bug fix: an error is signaled for tagbodies with duplicate tags
567 (thanks to Stephen Wilson)
568 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
569 * bug fix: Win32 port can now handle foreign code unwinding Lisp
570 stack frames from alien callbacks.
571 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
572 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
573 65k pages (thanks to David Woodhouse)
574 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
576 changes in sbcl-1.0.1 relative to sbcl-1.0:
577 * new platform: FreeBSD/x86-64, including support for threading.
578 * new feature: the compiler stores cross-referencing information
579 about function calls (who-calls), macroexpansion (who-macroexpands)
580 and special variables (who-binds, who-sets, who-references) for code
581 compiled with (< SPACE 3). This information is available through the
582 sb-introspect contrib.
583 * new feature: users may subclass SEQUENCE, and have instances of
584 these classes interoperate with standard Common Lisp functions if
585 a number of methods are defined. (This feature is experimental
586 and the interface subject to change based on feedback from SBCL
587 users and the general community)
588 * improvement: sb-sprof traces call stacks to an arbitrary depth on
589 x86/x86-64, rather than the previous fixed depth of 8
590 * improvement: another pthread back-end of mutex "pthread-futex".
591 * bug fix: non-ascii command-line arguments are processed correctly
592 (thanks to Yaroslav Kavenchuk)
593 * bug fix: non-required arguments were not passed correctly when a method
594 defined using DEFMETHOD was called from a mop-generated method using
595 CALL-NEXT-METHOD (reported by Pascal Costanza)
596 * bug fix: recursion is now permitted in accessors through
597 SLOT-UNBOUND. (reported by Pascal Costanza)
598 * bug fix: an error was signaled at startup if the HOME environment
599 variable was defined, but had an empty value (reported by Peter Van Eynde)
600 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
601 signal a PROGRAM-ERROR, not a BUG.
602 * bug fix: SB-POSIX:READDIR works when built with large file support.
603 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
605 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
606 for better automatic indentation support. (thanks to Matt Pillsbury)
607 * optimization: loading generic functions no longer takes O(n^2) time,
608 proportional to the amount of methods in the generic function
609 (reported by Todd Sabin and Jeremy Brown)
610 * optimization: the FIND and POSITION family of sequence functions
611 are significantly faster on arrays whose element types have been
613 * improvements to the Windows port:
614 ** Intermittent heap corruption problems have been fixed. (thanks
615 to Alastair Bridgewater)
616 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
618 ** Lisp is able to unwind foreign exception frames from alien
619 callbacks. (thanks to Alastair Bridgewater)
621 changes in sbcl-1.0 relative to sbcl-0.9.18:
622 * improvement: experimental support for threading on FreeBSD/x86.
623 (thanks to NIIMI Satoshi)
624 * improvement: runtime option --dynamic-space-size can be used
625 to set the size of the dynamic space reserved on startup.
626 * improvement: floating point modes in effect are now saved in
627 core, and restored on startup.
628 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
629 startup, not time since first call to GET-INTERNAL-REAL-TIME.
630 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
631 threads are not running after *SAVE-HOOKS* have run.
632 * improvement: writes to CLOS instance slots are type-checked in code
633 compiled with (SAFETY 3)
634 * improvement: floating-point exception handling on FreeBSD (thanks to
636 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
637 (thanks to Zach Beane)
638 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
640 * improvement: added support for the Shift-JIS external format.
641 (contributed by NIIMI Satoshi)
642 * improvement: callbacks are supported on Linux/PPC. (thanks to
644 * bug fix: compiler bug triggered by a (non-standard) VALUES
645 declaration in a LET* was fixed. (reported by Kaersten Poeck)
646 * bug fix: file compiler no longer confuses validated and already
647 dumped structurres. (reported by Kaersten Poeck)
648 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
649 fill-pointer signals a type-error as required. (thanks to
651 * bug fix: disassemly of funcallable instances works.
652 * bug fix: single stepping on PPC.
653 * bug fix: fix thread-safety problems in the type system (generally
654 manifesting as nonsensical errors like "STRING is a bad type specifier
655 for sequences" or "The value 1 is not of type FIXNUM")
656 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
657 (reported by Josip Gracin)
658 * bug fix: an error is signaled for attempts to displace arrays with
659 incompatible element types (thanks to Mario Mommer)
660 * bug fix: more correct handling of wide characters in debug info
661 (bug reported by Attila Lendvai and fixed by Juho Snellman)
662 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
663 and don't cause extra consing
664 * optimization: MAP and MAP-INTO are significantly faster on vectors
665 whose elements types have been declared.
666 * Improvements to SB-SPROF:
667 ** Support for allocation profiling
668 ** Reduced profiling overhead, especially for long profiling runs
669 * Improvements to the Windows port:
670 ** floating point exceptions are now reported correctly.
671 ** stack exhaustion detection works partially.
672 ** more accurate GET-INTERNAL-REAL-TIME.
673 ** better RUN-PROGRAM behaviour in case of errors in the child
675 ** PROBE-FILE now simplifies pathnames correctly.
676 ** DIRECTORY now works correctly with :WILD-INFERIORS.
677 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
678 1st 1970 (although time timezone data will be unreliable).
679 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
681 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
682 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
683 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
685 * bug fix: two potential GC deadlocks affecting threaded builds.
686 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
687 returns T (reported by Anton Kazennikov)
688 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
689 with non-variable places
690 * bug fix: remove a race condition in the setting of
691 funcallable-instance functions, this should make threaded CLOS
692 code more stable against memory faults.
693 * bug fix: corruption of specials when unbinding is interrupted by an
694 asynchronous unwind (reported by Hannu Koivisto)
695 * improvement: the debugger will now also display local variables that
696 are only used once, for code compiled with a DEBUG optimization quality
699 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
700 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
701 * incompatible change: External-format support for FFI calls. The
702 SB-ALIEN:C-STRING no longer implies an ASCII
703 external-format. Instead the string is subject to external-format
704 conversion. Additionally return values of type SB-ALIEN:C-STRING
705 are by default of type (SIMPLE-ARRAY CHARACTER), not
706 SIMPLE-BASE-STRING. For an alien type that behaves like the old
707 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
708 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
709 * incompatible change: SB-EXT package no longer contains the
710 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
711 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
713 * incompatible change: the single-stepper is no longer available
714 on Alpha, Mips and Sparc platforms.
715 * minor incompatible change: the direct superclasses of
716 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
717 not (STANDARD-OBJECT FUNCTION). This makes the
718 class-precedence-lists of GENERIC-FUNCTION and
719 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
721 * minor incompatible change: the default stream external format on
722 non-unicode SBCL is no longer detected from the locale, but is always
724 * new feature: Add a version of evaluator that uses an interpreter instead
725 of the compiler. EVAL still uses the compiler by default, to switch it
726 to use the interpreter, set the value of the variable
727 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
728 * minor incompatible change: the single-stepper REPL has been merged
729 with the normal debugger (see the "Stepping" heading of the debugger help
730 for more details). The debugger command STEP will no longer switch
731 to the single-stepper REPL.
732 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
733 for a type now works.
734 * bug fix: loading of default sysinit file works. (thanks to Leonid
736 * bug fix: better detection of circularities in the file-compiler.
737 (reported by Marco Monteiro)
738 * bug fix: the CL pathname functions now work with files that have
739 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
740 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
741 systems with Unix98 pty semantics.
742 * bug fix: ASDF-INSTALL will now work with bsd tar.
743 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
745 * bug fix: timers expiring in dead threads no longer cause a
746 type-error (reported by Paul "Nonny Mouse").
747 * bug fix: thanks to more lightweight single-stepper instrumentation,
748 code compiled with (DEBUG 3) will compile and execute significantly faster,
749 and will have more accurate type-inferencing than before
750 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
751 whose bindings are modified
752 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
753 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
754 ** replace dummy implementations of CL:MACHINE-INSTANCE and
755 CL:SOFTWARE-VERSION with proper version
757 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
758 * feature: implemented the READER-METHOD-CLASS and
759 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
760 as specified by AMOP.
761 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
763 * optimization: faster LOGCOUNT implementation on x86 and x86-64
764 (thanks to Lutz Euler)
765 * optimization: hashing of general arrays and vectors has been
766 improved. (reported by Any Fingerhut)
767 * enhancement: SB-INTROSPECT is now able to find definitions of
768 profiled functions. (thanks to Troels Henriksen)
769 * enhancement: compiler-macro expansion applies now to FUNCALL forms
770 as well, allowing compiler-macros for SETF-functions to expand.
771 * enhancement: step-instrumentation no longer wraps known
772 single-value functions with multiple-value context, allowing
773 better type inference.
774 * fixed bug #337: use of MAKE-METHOD in method combination now works
775 even in the presence of user-defined method classes. (reported by
776 Bruno Haible and Pascal Costanza)
777 * fixed bug #339(c): if there are applicable methods not part of any
778 long-form method-combination group, call INVALID-METHOD-ERROR.
779 (reported by Bruno Haible)
780 * fixed bug #361: the :FUNCTION initarg in the protocol for
781 initialization of methods can now be used to override
782 internally-produced optimized functions. (reported by Bruno
784 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
785 system-provided lambda expression no longer cause warnings about
786 unbound #:|pv-table| symbols.
787 * bug fix: improved the handling of type declarations and the
788 detection of violations for keyword arguments with non-constant
790 * bug fix: potentially erronous calls to PATHNAME and
791 MERGE-PATHNAMES were being flushed in some cases.
792 (reported by Richard Kreuter)
793 * bug fix: compiled calls to TYPEP were mishandling obsolete
794 instances. (reported by James Bielman and Attila Lendvai)
795 * bug fix: format strings with ~> without matching ~< no longer
796 trigger an AVER, but signal an understandable error instead.
797 (reported by Antonio Martinez)
798 * bug fix: specifying an output-file without a pathname-type for
799 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
800 of input-file instead of "fasl". (reported by Robert Dodier)
801 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
802 (reported by James Y Knight).
803 * bug fix: compiler-macros-function did not consider the environment
804 argument for shadowing by local functions.
805 * bug fix: compiler-macros expansion was inhibited by local INLINE
807 * bug fix: inline expansions of known functions were subject to
808 step-instrumentation in high debug policies, leading to problems
810 * bug fix: compiler failed to differentiate between different CONS
812 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
813 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
814 * bug fix: #368: incorrect use of expressed vs. upgraded array
816 * bug fix: #306a: more precise unions of array types.
817 * thread-safety improvements:
818 ** CONDITION-WAIT could return early on Linux, if the thread was
819 interrupted and subsequently continued with SIGCONT.
820 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
822 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
823 * added support for the ucs-2 external format. (contributed by Ivan
825 * minor incompatible change: pretty printing of objects of type
826 (cons symbol) is, in the default pprint-dispatch-table, now
827 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
829 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
830 called later in a class's lifetime, possibly as late as when the
831 first instance of the class is created. Previously,
832 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
833 class became finalizeable.
834 * fixed bug: FILE-POSITION sometimes returned inconsistent results
835 for multibyte external-format streams. (thanks to "vbzoli")
836 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
837 with :ALLOCATION :CLASS inherited from superclasses of the
839 * fixed bug: anonymous classes can now be created using the :NAME
840 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
841 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
842 * fixed bug: core-files saved with :EXECUTABLE T can again be
843 executed when SBCL_HOME isn't set. (reported by James Knight)
844 * fixed bug: toplevel LOCALLY forms with declarations could
845 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
846 * fixed bug: printing from several different threads using different
847 values of *print-case* could cause invalid output, due to
848 some internal special variables of the printer not being bound
849 thread-locally (reported by Max Mikhanosha)
850 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
851 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
852 (reported by Pascal Costanza)
853 * fixed bug: FINALIZE-INHERITANCE is called from
854 REINITIALIZE-INSTANCE on classes when the class has previously
855 been finalized, as required by AMOP.
856 * minor code generation optimizations:
857 ** better register allocation in CLOS dispatching functions
858 ** overflow detection when coercing signed bytes to fixnums on x86-64
859 ** is now implemented with one IMUL instruction instead of three shifts
860 ** more efficient bit-vector access on x86 and x86-64
861 ** more efficient access to raw structure slots on x86 and x86-64
862 * fixed some bugs revealed by Paul Dietz' test suite:
863 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
866 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
867 * feature: thread support on Solaris/x86, and experimental thread support
869 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
871 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
872 default method for SLOT-UNBOUND.
873 * minor incompatible change: prevent the user from specializing the
874 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
875 somewhat counter to the intent of the protocol, I (CSR) think, and
876 additionally it just doesn't work in SBCL as currently
877 implemented, thanks to optimizations (that are always valid for
878 the other three SLOT-VALUEish functions, but not for the setter).
879 * minor incompatible change: the :SB-LDB feature is now enabled by
880 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
881 the low-level debugger.
882 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
883 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
884 down to an SB-SYS:FD-STREAM, allowing direct communication in
885 more cases. Temporary files and pipes are still used as a fallback
887 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
888 on the non-reentrant C functions localtime(3) and gmtime(3).
889 * bug fix: global optimization policy was not visible in LOCALLY and
891 * bug fix: class objects can be used as specializers in methods.
892 (reported by Pascal Costanza)
893 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
894 with a type error. (reported by blitz_ on #lisp)
895 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
896 failed with a type error. (reported by Pascal Bourguignon)
897 * bug fix: merging pathnames against defaults with :DIRECTORY
898 starting with '(:RELATIVE :BACK) should preserve the :BACK.
899 (reported by James Y Knight)
900 * bug fix: saving large (>2GB) cores on x86-64 now works
901 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
902 constant index and a (simple-array (signed-byte 32)) array
903 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
904 error (patch by Robert J. Macomber)
905 * feature: TIME now displays the amount of run-time spent in GC
906 * bug fix: The debugger now does a better job of respecting
907 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
908 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
909 workaround for bug 403.)
910 * bug fix: floating point exception handling now works on FreeBSD
911 (thanks to NIIMI Satoshi)
912 * fixed some bugs revealed by Paul Dietz' test suite:
913 ** MISC.641: LET-conversion were not supposed to work in late
915 * improvements to the Win32/x86 port:
916 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
917 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
919 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
920 * new feature: source path information is generated for macro-expansion
921 errors for use in IDE's like Slime (thanks to Helmut Eller)
922 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
923 * bug fix: compiler does not loop forever on an invalid type in
925 * improvement: compilation of most CLOS applications is significantly
927 * optimization: added a limited bytecode compiler for simple toplevel
928 forms, speeding up compilation and FASL loading
929 * bug fix: the statistical profiler now properly distinguishes anonymous
932 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
933 * minor incompatible change: in sbcl-0.9.11 (but not earlier
934 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
935 into subclasses of the finalized class. Now user calls to
936 FINALIZE-INHERITANCE finalize just the one class, and calls by the
937 system return before any subclasses are finalized.
938 * minor incompatible change: The reader no longer ignores errors
939 regarding non-existent packages in #+ and #- feature tests.
940 * new feature: command line options --no-sysinit, --no-userinit to
941 inhibit loading the corresponding init files
942 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
943 for sending data through UDP sockets (thanks to François-René Rideau)
944 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
945 error is signalled instead (thanks to François-René Rideau)
946 * minor incompatible change: Error signalling behaviour of lexical
947 operations violating package locks has changed slightly. Refer to
948 documentation on package locks for details.
949 * enhancement: EVAL can process IF-expressions without resorting to the
951 * bug fix: Some operations involving SIN, COS, or TAN weren't being
952 constant-folded properly due to missing out-of-line functions.
953 (reported by Mika Pihlajamäki)
954 * bug fix: LISTEN sometimes returned T even in cases where no data was
955 immediately available from the stream
956 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
957 were reversed. (reported by Levente Mészáros)
958 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
959 when given a vector displaced to an adjustable array. (reported
961 * bug fix: derivation of float boundaries from numbers outside the
962 appropriate float range (reported by John Wiseman)
963 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
965 * bug fix: precomputing cache entries for generic functions with
966 some subclasses of specializers as yet invalid does not attempt to
967 fill a cache line with a negative offset. (reported by Levente
969 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
970 allow condition class objects as arguments to DOCUMENTATION and
971 (SETF DOCUMENTATION); only find and set documentation for
972 structure names for the STRUCTURE doc-type. (suggested by Gary
974 * improvements to the Win32/x86 port:
975 ** (user-homedir-pathname) and default initialization file
976 locations now know about the user's "Documents and Settings"
977 directory (thanks to Yaroslav Kavenchuk)
978 ** run-program is implemented (thanks to Mike Thomas)
979 ** sockets support (thanks to Timothy Ritchey)
980 ** better backtrace support (thanks to Alastair Bridgewater)
981 ** sb-grovel supported
982 ** asdf-install and sb-posix work somewhat
983 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
984 * improvements to the Solaris/x86 port:
985 ** works on Solaris 11/Solaris Express
986 ** floating-point exception handling support
987 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
988 * fixed some bugs revealed by Paul Dietz' test suite:
989 ** REMOVE-METHOD returns its generic function argument even when
990 the method is not one of the generic functions' methods.
991 ** objects declared as MEMBER types can be admissible arguments to
993 ** printing characters should simply be printed by the FORMAT ~:C
995 ** compiler failure when compiling functions with hairy constant
996 defaults for optional parameters.
997 ** compiler produces wrong code when MAYBE-INLINE-expanding a
998 function, which is already optimized.
1000 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1001 * new platform: experimental support for SBCL x86/Darwin, including
1002 MacOS X 10.4.5 on Intel.
1003 * new feature: Unicode character names are now known to the system
1004 (through CHAR-NAME and NAME-CHAR).
1005 * new feature: the filesystem location of SBCL's core file is
1006 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1007 * minor incompatible change: the contrib modules SB-POSIX and
1008 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1009 this change is to make it easier to distribute
1010 location-independent binaries.
1011 * bug fix: as implied by AMOP, standardized classes no longer have
1012 slots named by external symbols of public packages. (reported by
1014 * optimization: calling functions via constant symbols -- as in
1015 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1016 via the function object as in (FUNCALL #'FOO).
1017 * optimization: on x86 and x86-64, the calling convention for the
1018 case of dealing with an unknown number of values has been altered
1019 to be friendlier to the prediction heuristics implemented,
1020 particularly with reference to CALL and RET pairing. (thanks to
1021 Alastair Bridgewater)
1022 * optimization: on x86, the code for access to thread-local symbol
1023 values and binding/unbinding of thread-local symbols is smaller.
1024 (thanks to Alastair Bridgewater)
1025 * enhancement: CONSTANTP is now able to determine constantness of
1026 more complex forms, including calls to constant-foldable standardized
1027 functions and some special forms beyond QUOTE.
1028 * fixed bug: occasional GC crashes on Solaris/x86
1029 * optimization: x86-64 supports stack allocation of results of simple
1030 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1031 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1032 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1033 (thanks to James Knight)
1034 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1035 by the standard (thanks to Stephen Compall)
1037 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1038 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1039 be used for bundling the runtime and the core file into one
1040 executable binary. This feature is not currently supported on all SBCL
1041 platforms. (thanks to James Bielman and NIIMI Satoshi)
1042 * new feature: a generational or ephemeral garbage collector is now
1043 the default on the PowerPC platform (both Linux and Darwin). The
1044 old Cheney (stop and copy) collector is a build-time option.
1045 (thanks to Cyrus Harmon, after Raymond Toy)
1046 * minor incompatible change: the method by which SBCL finds its
1047 contributed modules has changed; it no longer relies on symbolic
1048 links from an $SBCL_HOME/systems directory, but searches directly
1049 in the subdirectories of $SBCL_HOME.
1050 * enhancement: the dynamic heap size on the Linux/PPC platform is
1051 markedly larger, even using the older Cheney garbage collector.
1052 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1053 certain complicated string types. (reported by Gary King)
1054 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1055 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1056 * fixed bug: DOCUMENTATION on structures no longer signals an
1057 error if no structure type of the right name exists. (reported by
1059 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1060 applicable non-standard primary or :AROUND method on
1061 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1062 no longer cause unbound variable errors. (reported by Kevin Reid)
1063 * fixed bug: metacircle resolution in cases where methods have slots
1064 added before the slots from STANDARD-METHOD. (reported by Jean
1066 * fixed bug: the Power PC assembly code for calling into lisp
1067 sometimes computed its return address wrongly (depending on the
1068 15th bit of the address, fixed at link time). This used to
1069 manifest itself in a segmentation violation while building PCL.
1070 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1071 many others over the years)
1072 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1073 the stack pointer (treated as a signed integer) is negative.
1074 (thanks to Peter van Eynde)
1076 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1077 * new platform: experimental support for the Windows operating
1078 system has been added. (thanks to Alastair Bridgewater)
1079 * fixed several bugs in and robustified the PPC FFI (including
1080 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1081 * bug fix: allow non-simple string symbol names (reported by Paul
1083 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1085 * bug fix: saving a core corrupted callbacks on x86/x86-64
1086 * bug fix: closed a loophole in metacircularity detection and
1087 grounding in the PCL implementation of CLOS.
1088 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1089 (reported by tomppa on #lisp)
1090 * bug fix: clear the direction flag on Lisp -> C transitions, as
1091 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1092 (reported by Andrej Grozin and Hendrik Maryns)
1093 * optimization: major improvements to GC efficiency on GENCGC platforms
1094 * optimization: faster implementation of EQUAL
1095 * optimization: emit more efficient opcodes for some common
1096 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1097 * optimization: several other minor code-generation improvements on x86-64
1098 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1100 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1101 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1102 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1103 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1104 by ANSI as a generic function, and for consistency (SETF
1105 GENERIC-FUNCTION-NAME) is treated likewise.
1106 * fixed bug #233.b: make constraint propagation notice when a variable
1107 value is changed after it is referenced but before it is used
1108 * fixed bug #296: no more arbitrary behaviour from filenames with
1109 odd characters as --load arguments. (reported by Adam Warner)
1110 * fixed bug #390: :CHARACTER-SET pathname components now work as
1111 expected. (reported by Tim Daly Jr)
1112 * fixed bug #391: complicated :TYPE intersections in slot
1113 definitions no longer cause an error in PCL internals.
1114 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1115 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1116 returning the number of octets which would be written to the
1117 file-stream. (thanks to Robert J. Macomber)
1118 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1119 arguments to the internal stream functions. (thanks to David
1121 * bug fix: stack exhaustion is now handled on the x86/SunOS
1122 platform. (thanks to Daisuke Homma)
1123 * bug fix: more accurate ROOM results on GENCGC platforms
1124 * optimization: improved type inference for arithmetic-for
1125 index variables in LOOP
1126 * optimization: faster floating-point SQRT on x86-64
1127 * fixed some bugs revealed by Paul Dietz' test suite:
1128 ** DOCUMENTATION returns NIL instead of "" for method combinations
1129 that don't have a docstring
1131 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1132 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1133 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1134 therefore are not customizeable by user code (as seems to be at
1135 least permitted and maybe required by AMOP). As a consolation,
1136 however, the SBCL implementation of these functions now calls
1137 REINITIALIZE-INSTANCE as specified by AMOP.
1138 * bug fix: slot-definition documentation is propagated as per ANSI
1139 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1140 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1141 argument, as do eslotd initialization methods. (from Pascal
1142 Costanza's "Closer" project)
1143 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1144 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1146 * bug fix: it is now possible to have more than one subclass of
1147 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1148 (reported by Bruno Haible, Pascal Costanza and others)
1149 * bug fix: the dependent update protocol now works for generic
1150 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1151 and Pascal Costanza)
1152 * bug fix: condition-class instances corresponding to
1153 DEFINE-CONDITION forms are now created eagerly. (reported by
1154 Kalle Olavi Niemitalo on comp.lang.lisp)
1155 * bug fix: floating point printing is more accurate in some
1156 circumstances. (thanks to Simon Alexander)
1157 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1158 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1159 * bug fix: callbacks on OS X now preserve stack-alignment.
1160 (thanks to Cyrus Harmon)
1161 * enhancement: the x86-64 disassembler is much better at
1162 disassembling SSE instructions. (thanks to Lutz Euler)
1163 * enhancement: CHECK-TYPE on a variable now allows type-inference
1164 (thanks to Vincent Arkesteijn)
1165 * optimization: improved performance of EUC-JP external format.
1166 (thanks to NIIMI Satoshi)
1167 * optimization: performance improvements to IO on file streams of
1168 :ELEMENT-TYPE CHARACTER
1169 * optimization: much faster memory allocation on x86-64
1170 * optimization: faster garbage collections (latency and throughput)
1172 * optimization: faster bignum subtraction and division on x86 and x86-64
1173 (thanks to Lutz Euler)
1174 * optimization: more accurate type-inference for loops that use a
1175 floating point index variable or a negative step.
1177 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1178 * bug fix: add a workaround to SBCL looping infinitely at startup on
1179 Linux kernels with apparently buggy implementations of personality().
1180 (thanks to Svein Ove Aas)
1181 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1182 (thanks to David Lichteblau)
1183 * bug fix: local bindings shadow global symbol macros for
1184 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1185 * optimization: non-open coded uses of numeric comparison operators
1186 (e.g. >) no longer cons when called with more than one parameter
1187 on platforms supporting dynamic-extent allocation.
1188 * enhancement: saving cores with foreign code loaded is now supported
1189 on MIPS/Linux in addition to the previously supported platforms.
1190 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1191 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1192 * bug fix: division by zero in sb-sprof when no samples were collected
1193 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1194 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1195 (thanks to Alexander Kjeldaas)
1196 * new feature: ignoring the return values of destructive functions
1197 like NREVERSE signals a compile-time style-warning.
1198 (thanks to Kevin Reid)
1199 * new feature: the alignment of alien structure fields can be
1200 explicitly specified. (thanks to Cyrus Harmon)
1201 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1202 and dump core on SIGQUIT
1204 ** incompatible change: threads do not inherit values of specials
1205 from their parents (see manual)
1206 ** bug fix: threads stacks belonging to dead threads are freed by the
1207 next exiting thread, no need to gc to collect thread stacks anymore
1208 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1209 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1210 ** bug fix: dynamic variable and thread start related gc lossage
1211 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1213 * fixed some bugs revealed by Paul Dietz' test suite:
1214 ** SUBTYPEP is slightly more accurate on heinously complicated
1215 CONS types where some of the members have uncertain (in the
1216 NIL, NIL sense) type relationships to each other.
1218 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1219 no longer purified unless :PURIFY T is explicitly specified.
1220 ** Non-purified cores are significantly smaller than before
1222 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1223 * new feature: timers based on Zach Beane's excellent timer package
1224 * added support for the following external formats: koi8-u,
1225 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1226 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1227 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1228 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1229 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1230 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1231 * incompatible change: a threaded SBCL will no longer revert to
1232 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1233 * bug fix: interrupts are disabled until startup is complete; no
1234 more sigsegvs when receiving a signal to soon
1235 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1237 * bug fix: add a workaround for the memory randomization features in
1238 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1239 workaround will only be in effect on systems with the proc filesystem
1241 * bug fix: printing objects of type HASH-TABLE signals a
1242 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1244 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1245 have been running for over 50 days. (reported by Gilbert Baumann)
1246 * bug fix: the logic for getting names of functions gets less
1247 confused when confronded with alternate-metaclass
1248 funcallable-instances. (reported by Cyrus Harmon)
1249 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1250 consistent, even on internal alternate-metaclass objects.
1251 * bug fix: flush closure information collected by physical
1252 environment analysis prepass before the main pass. (bug reported
1254 * bug fix: compiler pack phase does not modify a hashtable, which is
1255 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1256 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1257 may be expanded. (reported by Alexander <malishev> on c.l.l)
1258 * bug fix: no more style-warnings when compiling code using LAST
1259 under high-SPEED low-SPACE settings. (reported by David Wallin
1261 * fixed some incompatibilities between SBCL's MOP and the MOP
1263 ** the METAOBJECT class is now implemented;
1264 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1265 STANDARD-OBJECT, as required;
1266 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1267 now compatible (as required by VALIDATE-SUPERCLASS); there
1268 remains a consistency requirement over the presence or absence
1269 of the FUNCTION class in the superclass list of finalized
1270 classes; see the manual for more details;
1271 ** the :AROUND method for COMPUTE-SLOTS on
1272 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1273 requested slot ordering.
1275 ** bug fix: parent thread now can be gc'ed even with a live
1277 ** bug fix: binding a special with PROGV to no value is not
1278 the same as that symbol not having been bound (thanks to
1280 ** bug fix: binding specials is thread safe (thanks to
1282 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1283 and SB-SPROF do not die with 'no handler for signal XX in
1284 interrupt_handle_now(..)' anymore
1285 ** bug fix: WITH-TIMEOUT works with multiple running threads
1286 * fixed some bugs revealed by Paul Dietz' test suite:
1287 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1288 the :method-class keyword argument.
1290 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1291 * new port: the Solaris operating system on x86 processors is now
1292 mostly supported, though some rough edges in the environment
1293 remain. (thanks to Daisuke Homma)
1294 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1295 space for its heap. (thanks to Thiemo Seufer)
1296 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1297 closures. (thanks to Thiemo Seufer)
1298 * minor incompatible change: eof selects abort in the debugger.
1299 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1300 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1302 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1303 deprecated, and will go away in a future revision of SBCL.
1304 * minor incompatible change: GC-ON and GC-OFF are no longer
1305 implemented with a counter, it does not matter how many times gc
1306 is switched on or off
1307 * bug fix: discriminating functions for generic function classes
1308 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1309 make invalid assumptions about method precedence order. (reported
1311 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1312 platforms. (thanks to Thiemo Seufer)
1313 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1314 * bug fix: correct pathname printing with printer escaping is on.
1315 (thanks to Kevin Reid)
1316 * bug fix: complex VOP definitions in "user-space" no longer trigger
1317 package locks. (reported by Zach Beane)
1318 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1319 for funcallable-instance functions, and is no different from
1321 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1322 characters in the current readtable. (reported by Nicholas Neuss)
1323 * bug fix: SUBTYPEP on various CONS types returns more a more
1324 accurate acknowledgment of its certainty.
1325 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1326 lists in some cases. This partially fixes bug 384.
1327 * flush all standard streams before prompting in the REPL and the
1329 * bug fix: signal handling and triggering gc do not conflict
1330 directly anymore, in particular a high frequency sb-sprof does
1331 not prevent gc from running
1332 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1333 approximation for timezone and DST information between the
1334 universal time epoch and the smallest negative 32-bit time_t.
1335 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1336 year 1899 so long as the encoded time is non-negative.
1337 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1338 the correct answer for non-constant rotate amounts when performing
1339 an inline 32-bit rotation.
1341 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1342 there is only one thread in the session
1343 ** bug fix: memory leak for streams created in one thread and
1344 written to in another
1345 ** bug fix: lockup when compiled with gcc4
1346 ** bug fix: race that allows the gc to be triggered when gc is
1348 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1349 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1351 ** bug fix: finished threads are gc'ed properly
1352 * fixed some bugs revealed by Paul Dietz' test suite:
1353 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1354 has arguments with a different set of applicable methods from
1355 the orignal arguments.
1356 ** The type error thrown by MAP now has a correct expected-type
1358 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1359 name a compiled function.
1360 ** Three MISC tests where a large bignum was improperly coerced to
1361 a float that couldn't represent that bignum during type
1362 derivation were fixed.
1363 ** SUBTYPEP can now handle types involving the negation of a
1364 list-form FUNCTION type.
1365 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1366 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1367 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1369 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1370 * New feature: Experimental support for bivalent streams: streams
1371 opened with :element-type :default now allow character and binary
1372 (unsigned-byte 8) I/O
1373 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1374 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1376 * bug fix: no more highly sporadic "couldn't check whether ~S is
1377 readable" when reading a stream and an interrupt hits in the middle
1378 of a select system call
1379 * compiler better recognizes complex arrays (reported by Eduardo
1381 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1382 for variables on SBCL built with linkage-tables. (reported by Luis
1384 * various error reporting improvements.
1385 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1386 (Thanks to Thiemo Seufer)
1387 * bug fix: nested reader invokations work correctly
1388 * bug fix: it is possible to have simultaneous references to foreign
1389 code and foreign data with the same name.
1391 ** added x86-64 support
1392 ** incompatible change: the threading api now works with thread
1393 objects instead of thread ids
1394 ** bug fix: threads are protected from signals and interruption when
1395 starting up or going down
1396 ** bug fix: a race where an exiting thread could lose its stack to gc
1397 ** bug fix: don't halt on infinite error in threads if possible
1398 ** fixed numerous gc deadlocks introduced in the pthread merge
1399 ** bug fix: fixed thread safety issues in read and print
1400 ** bug fix: debugger doesn't hang on session lock if interrupted at
1401 an inappropriate moment
1402 ** bug fix: run-program is now thread safe(r)
1403 ** bug fix: inner with-recursive-lock no longer releases the mutex
1404 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1405 * fixed some bugs revealed by Paul Dietz' test suite:
1406 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1407 EXPECTED-TYPE slots filled.
1408 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1410 ** MAKE-CONDITION accepts classes as type-designators.
1411 ** COMPILE may never return NIL.
1412 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1413 range before calling Unix time functions
1415 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1416 * numerous signal handling fixes to increase stability
1417 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1418 * minor incompatible change: we now correctly canonize default
1419 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1420 as per the MOP, rather than the historical (INITARG INITFUNCTION
1421 INITFORM). (reported by Bruno Haible)
1422 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1423 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1424 for more information.
1425 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1426 pathname is a directory pathname.
1427 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1428 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1430 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1431 :METACLASS argument in addition to a class name. (reported by
1432 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1433 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1434 (reported by Bruno Haible; patch from Gerd Moellmann)
1435 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1437 * bug fix: more cleanups to the floating point exception handling on
1438 x86-64 (thanks to James Knight)
1439 * bug fix: the compiler does not try to constant fold calls of
1440 COERCE and BIT-* functions when they return freshly constructed
1441 objects. (reported by Nikodemus Siivola and Paul Dietz)
1442 * optimization: improved the MIPS versions of generic (in the
1443 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1444 * optimization: direct conversion of (unsigned-byte 32) to floats on
1445 the PowerPC platform.
1446 * optimization: structure instances with raw slots now use less
1447 memory, and probably show better memory locality. (thanks to
1449 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1450 are treated as requests for stack allocation on the x86-64,
1451 Alpha-32, PPC and SPARC platforms.
1452 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1453 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1454 (thanks to Tony Martinez)
1456 ** gcing a dead thread can no longer lead to lockups
1457 ** threads block signals until they are set up properly
1458 ** errno is no longer shared by threads
1459 ** interrupt-thread restores the eflags register on x86
1460 ** fixed some lockups due to gc/thread interaction
1461 * fixed some bugs revealed by Paul Dietz' test suite:
1462 ** invalid dotted lists no longer raise a read error when
1463 *READ-SUPPRESS* is T
1464 ** signal an error if a symbol that names a declaration is used
1465 as the name of a type, or vice versa
1466 ** allow using the (declare (typespec var*)) abbreviation for
1467 (declare (type typespec var*)) with all type specifiers
1468 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1469 ** always return NIL from PPRINT-POP when OBJECT is NIL
1470 ** don't signal errors when pretty-printing malformed LABELS,
1471 FLET or MACROLET forms
1472 ** declarations in a DOLIST body are also in scope for the
1474 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1476 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1479 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1480 * fixed cross-compiler leakages that prevented building a 32-bit
1481 target with a 64-bit host compiler.
1482 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1483 opened with :IF-EXISTS :OVERWRITE.
1484 * fixed bug 281: error for an invalid qualifier in a short-form method
1485 combination method is not signalled until the faulty method is called.
1486 * bug fix: iteration variable type inferrer failed to deal with open
1487 intervals. (reported by Alan Shields)
1488 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1489 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1490 or more runtime options were provided to the sbcl binary.
1491 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1492 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1493 * optimization: there's now a fast-path for fixnum arguments in the
1494 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1495 * optimization: the code generated on x86-64 is more compact thanks
1496 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1497 * bug fix: floating-point exception handling now partly works on
1498 x86-64. (Thanks to James Knight)
1499 * improvement to the MIPS backend from Thiemo Seufer: C-style
1500 64-bit long long arguments and return values to alien functions
1502 * fixed some bugs revealed by Paul Dietz' test suite:
1503 ** the type-error signalled from WARN has a filled-in DATUM slot.
1504 ** the type-error required when a stream is not associated with
1505 a file has the stream as its datum.
1506 ** type-errors on single-floats on x86-64 no longer have
1507 :INVALID-OBJECT as the datum
1508 ** the type-errors signalled for invalid function names now have
1509 a correct expected type
1510 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1511 for structures defined with a :TYPE. Documentation strings for
1512 typed structures are no longer immediately discarded
1513 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1514 broadcast streams and synonym streams. FILE-LENGTH now also works
1515 on broadcast streams.
1517 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1518 * incompatible change: the --noprogrammer option, deprecated since
1519 version 0.7.5, has been removed. Please use the equivalent
1520 --disable-debugger option instead.
1521 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1523 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1524 inoperational for a while now) has been completely removed.
1525 * null lexical environments are now printed as #<NULL-LEXENV>,
1526 significantly reducing the amount of clutter in typical backtraces.
1527 * documentation on weak pointers, finalization, and after GC hooks
1528 has been added to the manual.
1529 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1530 as other specialized array types, is much faster. SUBSEQ and
1531 COPY-SEQ on such arrays have also been sped up.
1532 * optimization: EQL is now more efficient when at least other argument
1533 is known to be of type (OR FIXNUM (NOT NUMBER)).
1534 * fixed bug: the runtime is now less vulnerable to changes in the
1535 size of the SBCL object on OS X, and virtual memory is reserved for
1536 all spaces that need to be at a fixed address.
1537 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1538 * fixed bug: finalizers and after GC hooks that cause consing are now
1540 * fixed bug: compiler error messages and summaries are now printed to
1541 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1542 * fixed inference of the upper bound of an iteration variable.
1543 (reported by Rajat Datta).
1544 * fixed bug 211e: calling local functions with duplicated constant
1545 keyword argument no longer causes a bogus style warning about an
1547 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1548 ftype declaration to be disregarded. (reported by Dave Roberts)
1549 * fixed bug 373: caused by erronous compilation of references to alien
1550 variables in the runtime on ppc/darwin.
1551 * fixed bug 376: CONJUGATE type deriver.
1552 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1553 (reported by Baughn on #lisp)
1554 * a cross-compiler bug on non-x86 platforms has been identified and
1555 fixed. (thanks to Bruno Haible)
1556 * improvements to the MIPS runtime code for increased stability.
1557 (thanks to Thiemo Seufer)
1558 * increased the maximimum compact environment size to allow
1559 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1560 * partial workaround for bug 135: don't name the function we're
1561 calling for hairy cases of EVAL, so as not to accumulate one environment
1562 entry per call to EVAL. (reported by Kevin Reid)
1563 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1564 * optimization: type testing for non-vector arrays should be faster.
1565 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1567 * bug fix: setting 31st element of a bit vector to zero did not work
1569 * bug fix: redefining a class definition which failed due to a
1570 previous accessor / function clash now works (but see BUGS entry
1571 #380 for more problems in this area). (thanks to Zach Beane)
1572 * the long form of DEFINE-METHOD-COMBINATION disables method group
1573 checking when given a single method group with pattern *. (thanks
1575 * on x86 compiler supports stack allocation of results of simple
1576 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1577 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1578 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1579 * fixed some bugs related to Unicode integration:
1580 ** the restarts for recovering from input and output encoding
1581 errors only appear when there is in fact such an error to
1583 * fixed some bugs revealed by Paul Dietz' test suite:
1584 ** MISC.549 and similar: late transformation of unsafe type
1585 assertions into derived types caused unexpected code
1587 ** SCALE-FLOAT type deriver is less wrong.
1588 ** type derivers for EXP, LOG and similar functions compute result
1589 types for complex arguments better.
1590 ** (MISC.563) CONJUGATE type deriver works for very restricted
1592 ** out-of-line type testers for character strings are available.
1593 ** EQUAL compiler transform understands specialness of objects
1595 ** accessing double-floats stored on the stack now works on x86-64.
1596 ** debugger internals could sometimes create invalid lispobjs,
1597 resulting in GC crashes.
1598 ** MISC.548: type check weakening can convert required type into
1600 ** initialization forms for bindings are not in scope of free special
1603 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1604 * incompatible change: thread support for non-NPTL systems has
1605 been removed - locking is buggy and unreliable. A threaded
1606 SBCL build will now warn at startup and refuse to create
1607 new threads, unless futex support is detected in the kernel
1608 * incompatible change: the top level REPL now has only an ABORT
1609 restart associated with it, not TOPLEVEL and ABORT as it used to.
1610 TOP and TOPLEVEL are now available as debugger commands for
1611 returning to the top level.
1612 * incompatible change: forms evaluated in the REPL now use the
1613 global optimization policy.
1614 * incompatible change: user- and system-initialization files are
1615 no longer processed with LOAD, but by READ and EVAL; hence the
1616 global optimization policy, startup package, readtable, etc,
1618 * The .fasl file format number has been incremented because of
1619 various incompatible changes.
1620 * internal entry point details and argument counts no longer appear
1621 in backtraces unless explicitly requested by setting
1622 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1623 * built-in and standard functions no longer have names like "top
1624 level local call to FOO".
1625 * fixed bug 32: functions defined in non-null lexical environments
1626 now have more legible printed representation
1627 * fixed bug 33: functions defined in non-null lexical environemnts
1628 are now more amenable to inspection by INSPECT.
1629 * workaround for bug 354: XEPs no longer appear in backtraces unless
1630 explicitly requested.
1631 * fixed bug: receiving the signal which results from attempting to
1632 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1633 SIGBUS on other *BSDs) on architectures where the C stack is also
1634 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1635 to debugger confusion or wild instability. (reported by Cheuksan
1637 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1638 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1639 (reported by Lutz Euler)
1640 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1641 compiled code. (reported by Paul Dietz)
1642 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1643 specialized parameter to a method to something that is not TYPEP
1644 the specializer is now possible.
1645 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1646 face of package deletion.
1647 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1648 pathname. (reported by David Tolpin on c.l.l)
1649 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1650 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1651 STRING-STREAM and FILE-STREAM, respectively.
1652 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1653 than previously; in addition, its test suite is now run on build.
1654 (thanks to Luke Gorrie)
1655 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1656 * optimization: added a immediate representation for single-floats
1658 * fixed some bugs revealed by Paul Dietz' test suite:
1659 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1660 ** attempting to create a package with a colliding nickname causes
1661 correctable errors to be signalled.
1662 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1663 ** several x86-64 backend bugs related to sign-extension of immediate
1666 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1667 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1668 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1670 * fixed loading of multiply forward-referenced layouts.
1671 (thanks to Cheuksan Wang)
1672 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1673 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1674 (see COMPILE-FILE documentation for details of :PRINT option)
1675 * fixed bugs 19 and 317: fixed-format floating point printing is
1676 more accurate. This also fixes a bug reported by Adam Warner
1677 related to the ~@F format directive.
1678 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1680 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1681 dispatch table if the from-char is a dispatch macro character.
1682 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1683 by the wordsize no longer ignore the last word. (reported by Lutz
1685 * fixed bug in type checking of dynamic-extent variables. (reported
1687 * optimization: sequence traversal functions use their freedom to
1688 coerce function designators to functions.
1689 * optimization: code with many calls to CLOS methods specialized on
1690 CLOS classes has had redundant type checks removed.
1691 * optimization: added declarations to speed up operations that access
1692 the internal character database (for example STRING-UPCASE)
1693 * optimization: comparison operations between floats and sufficiently small
1694 fixnums no longer create extra rationals
1695 * fixed some bugs related to Unicode integration:
1696 ** portions of multibyte characters at the end of buffers for
1697 character-based file input are correctly transferred to the
1698 start of the buffer at the next read.
1699 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1700 passing it through to OPEN.
1701 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1702 argument given to internal calls to OPEN.
1703 ** the built-in comment readers (introduced by character sequences
1704 ";" and "#|") are more forgiving to encoding errors; they will
1705 STYLE-WARN and then attempt to resync the stream at a character
1706 boundary. (thanks to Teemu Kalvas)
1707 * fixed some bugs in the x86-64 port:
1708 ** Negative short int return values from c-calls are sign-extended
1710 ** The stack is aligned to 16-bytes for c-calls, as required by
1711 the ABI. (thanks to Cheuksan Wang)
1712 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1713 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1714 for enums. (thanks to Vincent Arkesteijn)
1715 ** Multiple small optimizations and bugfixes for floating point
1717 * fixed some bugs revealed by Paul Dietz' test suite:
1718 ** Space, Tab, Linefeed, Return and Page have the invalid
1719 secondary constituent character trait.
1720 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1722 ** WITH-INPUT-FROM-STRING should only update the index place on
1724 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1725 works more reliably.
1726 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1727 with one-dimensional arrays with fill pointers.
1728 ** TYPE-OF failed on a complex with an integer realpart and a
1730 ** compiler failure during type inference for the code of form
1731 (IF (EQL X (THE ...)) ...) (MISC.535).
1733 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1734 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1735 platform. The port supports 61-bit fixnums, large memory spaces
1736 and reloading shared object files.
1737 * enhancement: saving cores with foreign code loaded is now
1738 supported on ppc/Darwin in addition to the previously supported
1740 * enhancement: the statistical profiler now walks deeper into the
1741 call stack for more meaningful call-graphs and accrued time
1742 reports (x86/x86-64 only). It also now reports time spent in
1744 * enhancement: it is now possible to trace most individual methods
1745 of a generic function in addition to tracing the generic function
1747 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1749 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1750 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1751 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1752 produces an error. (thanks to Vincent Arkesteijn)
1753 * bug fix: NAMESTRING on pathnames with :WILD components in their
1754 directories works correctly. (thanks to Artem V. Andreev)
1755 * fixed bug 125: compiler preserves identity of closures. (reported
1757 * bug fixed: functions with &REST arguments sometimes failed with
1758 "Undefined function" when compiled with (DEBUG 3). (reported by
1760 * bug fix: overflow during compiling of setting element of a bit
1761 vector with constant index and value. (reported by Timmy Douglas)
1762 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1763 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1764 * on x86 compiler partially supports stack allocation of dynamic-extent
1766 * GO and RETURN-FROM do not check the extent of their exit points
1767 when compiled with SAFETY 0.
1768 * fixed some bugs related to Unicode integration:
1769 ** encoding and decoding errors are now much more robustly
1770 handled; it should now be possible to recover even from invalid
1771 input or output to the terminal. (thanks to Teemu Kalvas)
1772 ** provided a first cut at implementing STRING-TO-OCTETS and
1773 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1774 ** altered the SB-MD5 contributed module slightly, changing the
1775 interface just enough for it to be supportable for builds where
1776 lisp characters are not eight bits.
1777 * fixed some bugs revealed by Paul Dietz' test suite:
1778 ** the FORMATTER-generated functions for ~V[ conditionals require
1779 the correct number of arguments.
1780 ** READ-FROM-STRING returns the mandated second value when applied
1781 to displaced strings.
1782 ** the #\Rubout and #\Backspace characters are treated as invalid
1783 constituent characters by the tokenizer.
1785 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1786 * new feature: reloading changed shared object files with
1787 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1788 * new feature: references to foreign variables and functions
1789 can now be compiled and loaded before the corresponding shared
1790 object file is loaded, as long as the foreign definitions are
1791 available at runtime.
1792 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1793 supported. (thanks to Dan Debertin)
1794 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1795 just keywords. (thanks to Vincent Arkesteijn)
1796 * fixed bug #331: structure-class instances corresponding to
1797 DEFSTRUCT forms are now created eagerly.
1798 * fixed bug #345: backtraces from calls to undefined functions work
1799 on x86 as well. Related bug #61 is now also partially fixed on x86
1800 (backtraces from throws to unknown catch tags.)
1801 * bug fix: lambda-list parsing is now stricter vrt. order and number
1802 of lambda-list keywords.
1803 * bug fix: as specified by AMOP, an error is signalled if a
1804 class-option appears multiple times in a DEFCLASS form. (reported
1806 * bug fix: RANDOM can be compiled when the compiler derives the type
1807 of its numeric argument as a disjoint set of small integers.
1808 (reported by Paul Dietz)
1809 * bug fix: starting a core saved with shared objects loaded when
1810 those objects are not available no longer causes threaded SBCL to
1811 hang. (reported by Sean Ross)
1812 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1814 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1815 argument, which must always be NIL. (reported by Kalle Niemitalo)
1816 * bug fix: printing 1.0d+23 no longer results in an error.
1817 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1818 * bug fix: structure slot setters preserve evaluation order. (thanks
1820 * fixed some bugs related to Unicode integration:
1821 ** RUN-PROGRAM can allow its child to take input from a Lisp
1822 stream. (reported by Stefan Scholl)
1823 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1824 locales. (reported by Ken Causey)
1825 * fixed some bugs revealed by Paul Dietz' test suite:
1826 ** INCF, DECF and REMF evaluate their place form as specified in
1828 ** FORMATTER expands ~{ ~} iteration directives with V or #
1829 parameters correctly.
1830 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1831 consequent uses no arguments correctly.
1832 ** the system has a partial understanding of the (COMPLEX RATIO)
1834 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1837 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1838 * new feature: a build-time option (controlled by the :SB-UNICODE
1839 keyword feature, enabled by default) for building the system with
1840 support for the entire 21-bit character space defined by the
1842 * new feature: the system now has rudimentary external-format
1843 support; the primary user-visible change at this time is that
1844 characters with the high bit set (such as lower-case-e-acute) will
1845 print correctly to a terminal in a UTF-8 environment.
1846 * minor incompatible change: BASE-CHAR no longer names a class;
1847 however, CHARACTER continues to do so, as required by ANSI.
1848 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1849 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1850 has been moved to the SB-EXT package (temporarily re-exported from
1852 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1853 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1855 * fixed bug #351: better error-handlind and reporting for malformed
1857 * fixed bug #350: bignum-printing is now more memory-efficient,
1858 allowing printing of very large bignums, eg. (expt 2 10000000).
1859 (reported by Bruno Haible)
1860 * fixed bug #302: better primitive-type selection for intersection
1862 * fixed bug #308: non-graphic characters now all have names, as
1863 required. (reported by Bruno Haible)
1864 * bug fix: redefining a class with different superclasses now correctly
1865 removes it from the direct-subclasses of its previous superclasses.
1866 (reported by David Morse)
1867 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1868 new value now works. (reported by Bruno Haible)
1869 * bug fix: correct canonicalization of multiple non-standard slot
1870 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1871 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1872 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1873 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1874 now exists, an signals an error.
1875 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1876 during macroexpansion for non-iteration variables as well. (reported
1877 by Bruno Haible for CMUCL)
1878 * bug fix: Cyclic structures and unprintable objects in compiler
1879 messages no longer cause errors. (reported by Bruno Haible)
1880 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1881 and READ-FROM-STRING all now return a primary value of NIL if
1882 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1883 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1884 (reported by Bruno Haible for CMUCL)
1885 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1886 do not designate a condition. (reported by Bruno Haible for
1888 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1889 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1890 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1891 permitting resolution in favour of any of the conflicting symbols.
1892 (reported by Bruno Haible for CMUCL)
1893 * bug fix: EQUAL compiler optimizations is less aggressive on
1894 strings which can potentially compare true despite having distinct
1895 specialized array element types.
1896 * bug fix: unit enumerations can be defined without dividing by
1897 zero. (thanks to Vincent Arkesteijn)
1898 * FORMAT compile-time argument count checking has been enhanced.
1899 (report from Bruno Haible for CMUCL)
1900 * a partial workaround for the bug 262: the compiler does not try to
1901 inline-expand a local function doing RETURN-FROM from a deleted
1902 BLOCK. (thanks to Peter Denno for the bug report and to David
1903 Wragg for the simple test case)
1904 * fixed some bugs revealed by Paul Dietz' test suite:
1905 ** RENAME-PACKAGE allows all package designators as new package
1907 ** constraint propagation and lambda variable substitution are
1908 more cautious in dealing with partially deleted code.
1909 ** compiler sometimes lost reoptimization passes.
1910 ** CERROR, given a condition as condition designator, passes its
1911 remaining arguments to the continue format control without
1913 ** Case-altering FORMAT directives work correctly on non-ASCII
1915 ** The REAL type specifier handles bounds outside the floating
1916 point ranges without signalling FLOATING-POINT-OVERFLOW.
1917 ** Functions with IR1-transformations can create intercomponent
1918 references to global functions.
1919 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1921 ** FORMAT ~R treats a nil value for its first parameter correctly.
1923 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1924 * enhancement: saving cores with foreign code loaded is now
1925 supported on x86/NetBSD and sparc/Linux in addition to the previously
1926 supported platforms.
1927 * bug fix: on some platforms repeated installations caused multiple
1928 copies of HTML documentation to be installed -- should not happen
1929 any more. (reported by Stefan Scholl)
1930 * bug fix: parsing self-recursive alien record types multiple times
1931 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1932 original patch by Helmut Eller for CMUCL)
1933 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1934 (thanks to Richard Kreuter)
1935 * bug fix: defining classes whose accessors are methods on existing
1936 generic functions in other (locked) packages no longer signals
1937 bogus package lock violations. (reported by François-René Rideau)
1938 * bug fix: special variables as DEFMETHOD parameters no longer have
1939 associated bogus type declarations. (reported by David Wragg and
1941 * bug fix: read-write consistency on streams of element-type
1942 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1943 * bug fix: redefiniton of the only method of a generic function with
1944 no DEFGENERIC no longer emits a full WARNING. In addition,
1945 redefinition of generic functions with no DEFGENERIC to an
1946 incompatible lambda list now signals an error. (thanks to Zach
1948 * bug fix: DEFGENERIC now works even when there's a function of the
1949 same name in an enclosing lexical environment. (thanks to Zach
1951 * fixed compiler failure, caused by instrumenting code during
1952 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1953 * optimization: added loop analysis and improved register allocation
1954 * fixed some bugs revealed by Paul Dietz' test suite:
1955 ** POSITION on displaced vectors with non-zero displacement
1956 returns the right answer.
1957 ** (SIMPLE-STRING) is a valid type specifier for sequence
1959 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1961 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1962 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1964 ** The justification version of the FORMAT ~< directive treats
1965 non-zero minpad parameter correctly.
1967 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1968 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1969 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1970 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1971 the supported interface.
1972 * new feature: Single-stepping of code compiled with DEBUG 2 or
1973 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1974 * new feature: saving cores with foreign code loaded is now
1975 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1976 Timothy Moore's work for CMUCL)
1977 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1978 parameters to * instead of NIL if no initform is supplied.
1979 (reported by Johan Bockgaard)
1980 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1981 correctly when FUNCALL appears as the car of the form. Note:
1982 despite this FUNCALL forms are not currently subject to
1983 compiler-macro expansion. (port of Raymond Toy's fix for the
1984 same from CMUCL, reported by Johan Bockgaard)
1985 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1986 (thanks for Teemu Kalvas)
1987 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1988 case instead of terminating the iteration (thanks for Julian
1989 Squires, Sean Champ and Raymond Toy)
1990 * bug fix: incorrect expansion of defgeneric that caused a style
1991 warning. (thanks for Zach Beane)
1992 * bug fix: slot accessor effective method computation works properly
1993 for classes with multiple non-standard applicable methods on
1994 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1995 * on x86 compiler supports stack allocation of results of LIST and
1996 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1997 CMUCL implementation by Gerd Moellmann)
1998 * fixed some bugs revealed by Paul Dietz' test suite:
1999 ** FORMAT strings with both the ~<~:;~> form of the justification
2000 directive and pretty-printing directives cause an error.
2002 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2003 * incompatible change: the internal functions
2004 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2005 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2006 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2007 instead of the old functions.
2008 * new feature: on platforms where "dladdr" is available foreign
2009 function names now appear in backtraces. (based on Helmut Eller's
2011 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2012 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2014 * documentation: ASDF manual is now distributed with SBCL.
2015 * bug fix: SBCL can now load its contributed modules with REQUIRE
2016 even if the system-provided entries have been removed from
2017 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2019 * bug fix: backtraces involving undefined functions or assembly
2020 routines are more informative. (thanks to Brian Downing)
2021 * bug fix: mutually referent alien structures now work correctly.
2022 (reported by Rick Taube)
2023 * bug fix: structures defined by WITH-ALIEN can be referred to
2024 within other definitions in the same WITH-ALIEN.
2025 * bug fix: division operators (MOD, TRUNCATE and the like) with
2026 constant zero divisors and integer dividends no longer generate
2028 * bug fix: provide default methods for INPUT-STREAM-P and
2029 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2030 * bug fix: improve the detection and resolution of MOP metacycles
2031 (where effective method computation depends on the generic function
2032 having its effective method computed). (reported by Bruno Haible)
2033 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2034 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2035 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2036 * bug fix: pathnames with relative directory components can be
2037 represented relative to default pathnames.
2038 * optimization: in taking the GCD of bignums, reduce the two bignums
2039 to approximately the same size (using Euclid's algorithm) before
2040 applying the more sophisticated binary GCD. (thanks to Juho
2042 * optimization: COUNT on bitvectors now operates word-at-a-time.
2043 * optimization: ASH with a positive, but not necessarily constant,
2044 (leftwards) shift, when in a modular context, is compiled to a
2046 * fixed some bugs revealed by Paul Dietz' test suite:
2047 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2048 if the corresponding argument is NIL.
2049 ** FORMAT directives accepting a minpad parameter treat negative
2051 ** literal commas as character parameters to FORMAT directives are
2053 ** literal spaces directly after ~<Newline> directives within a
2054 format-logical-block (~:< ~@:>) do not induce :FILL-style
2055 conditional newlines.
2056 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2057 as specified: it no longer includes conditional newlines).
2058 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2060 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2061 :PER-LINE-PREFIX argument does not evaluate to a string.
2063 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2064 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2065 the manual for details; for now, package locks can be disabled by
2066 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2067 if no major problems are found then it is likely that they will be
2068 compiled in unconditionally.
2069 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2070 unsupported operators on all platforms. To load a shared library
2071 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2072 object file, link it into a shared library outside of SBCL and
2073 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2075 * bug workaround: ROOM T and the GC were not working together
2076 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2077 reporting facility aren't preserved. That reporting has been
2078 disabled, so now until and unless someone figures out how to make it
2079 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2080 * minor incompatible change: as threatened around sbcl-0.8.0, the
2081 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2082 The symbols are also exported from SB-PCL for backwards
2083 compatibility, but more so than before SB-PCL should be treated as
2084 an implementation-internal package.
2085 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2087 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2088 It is known as of this release not to work on the Alpha, however.
2089 * fixed bug #167: errors signalled due to illegal syntax in method
2090 bodies are now more legible.
2091 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2092 designators and can hence be used with TYPEP.
2093 * fixed bug #333: CHECK-TYPE now ensures that the type error
2094 signalled, if any, has the right object to be accessed by
2095 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2096 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2097 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2099 * fixed bug #334: programmatic addition of slots using specialized
2100 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2101 and :ALLOCATION :CLASS slots.
2102 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2103 float exponents. (rereported by Peter Seibel)
2104 * fixed a bug: #\Space (and other whitespace characters) are no
2105 longer considered to be macro characters in standard syntax by
2106 GET-MACRO-CHARACTER.
2107 * fixed bug: initialization of condition class metaobjects no longer
2108 causes an instance of the condition to be created. (reported by
2110 * fixed bug: it is now possible to have slots such that
2111 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2112 description is neither :INSTANCE nor :CLASS.
2113 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2114 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2115 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2116 non-standard SLOT-DEFINITION-ALLOCATION values.
2117 * fixed bugs: various race conditions handling exiting threads.
2118 CL-PPCRE's thread torture test now passes.
2119 * fixed arguably-a-bug: GC time proportional to number of threads in
2120 system even when most of them are idle
2121 * optimization: improved performance of BIT and SBIT on bit-vectors.
2122 * .fasl file incompatibility: The fasl file version number has
2123 been incremented because of changes associated with package locks.
2125 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2126 * minor incompatible change: the system no longer provides
2127 optimization hints (conditions which are TYPEP
2128 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2129 mode; these hints are still emitted when higher SPEED optimization
2131 * new contrib module: a sampling profiler (profiling by statistical
2132 sampling, rather than by instrumenting functions) is available as
2133 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2134 * the behaviour of the standard function ED is now customizeable by
2135 third parties through a hook variable: see ED's documentation
2136 string for information on the protocol.
2137 * the compiler no longer emits efficiency notes for (FUNCALL X)
2138 when the type of X is uncertain under default optimization
2140 * fixed bug 276: mutating a binding of a specialized parameter to a
2141 method to something that is not TYPEP the specializer is now
2143 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2144 exhibit the required behaviour on the x86 platform. (thanks to
2145 Peter van Eynde, Eric Marsden and Bruno Haible)
2146 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2147 even for difficult arguments. (reported by Peter Graves)
2148 * fixed bug 141a: the backquote printer now descends quoted
2150 * fixed another bug in backquote printing: no more destructive
2151 modification of the form's list structure. (reported by Brian
2153 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2154 move between its address being taken and the call to
2155 interrupt_thread, fixing a crashing race condition.
2156 * the SB-POSIX contrib implementation has been adjusted so that it
2157 no longer exhibits ridiculously poor performance when constructing
2158 instances corresponding to C structs.
2160 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2161 * minor incompatible change: the sb-grovel contrib now treats C
2162 structures as alien (in the sense of SB-ALIEN) objects rather than
2163 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2164 has implications for memory management of client code
2165 (sb-grovel-returned objects must now be manually managed) and for
2166 type safety (alien objects now have full types).
2167 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2168 used to control emission of compiler diagnostics, rather than the
2169 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2170 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2171 quality should be considered deprecated.
2172 * install.sh now installs the user manual as well
2173 * (not quite a new documentable feature, but worth considering in
2174 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2175 beginnings of a semantically meaningful condition hierarchy is
2176 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2178 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2179 designator as the defaults argument.
2180 * fixed bug: Displaced arrays whose displaced-to array has become
2181 too small now cause ARRAY-DIMENSION to signal an error, providing
2182 for safer bounds-checking. (reported by Bruno Haible)
2183 * fixed bug: DEFCLASS slot definitions with identical :READER and
2184 :WRITER names now signal a reasonable error. (reported by Thomas
2186 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2188 * fixed bug: Invalid slot specification errors now print correctly.
2189 (thanks to Zach Beane)
2190 * fixed bug 320: Shared to local slot value transfers in class
2191 redefinitions now happen corrently from superclasses as
2192 well. (reported by Bruno Haible)
2193 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2195 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2196 improper lists correctly. (reported by Bruno Haible)
2197 * fixed bug 313: source-transform for <fun-name> was erroneously
2198 applied to a call of a value of a variable with name <fun-name>.
2199 (reported by Antonio Menezes Leitao)
2200 * fixed bug 307: The obsolete instance protocol ensures that
2201 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2202 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2203 remove dead unknown-values globs from the middle of the stack.
2204 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2205 easier to resume long computations after using *BREAK-ON-SIGNALS*
2206 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2207 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2208 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2209 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2210 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2211 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2213 * fixed a (fairly theoretical) bug in string printing: if
2214 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2215 does not have array-element-type equal to the most general string
2217 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2218 function name. (thanks to Nikodemus Siivola)
2219 * fixed bug: compile-time format string checker failed on
2220 non-closed ~{. (reported by Thomas F Burdick)
2221 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2222 don't warn on legal loop code involving more than one aggregate
2223 boolean. (thanks to Nikodemus Siivola)
2224 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2225 in keyword position to MAKE-PACKAGE was not being reported as
2227 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2228 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2230 * fixed bugs 280 and 312: the checking for multiple definitions in a
2231 file is less likely to become confused by uses of inline
2233 * fixed bug: the #S reader macro performs the keyword coercion
2234 specified for slot names. (reported by Kalle Niemitalo)
2235 * fixed bug: lambda lists may contain symbols whose names start with
2236 & but are not lambda-list-keywords; their occurrence triggers a
2238 * fixed bug 321: define-method-combination argument lambda lists do
2239 not cause invalid code to be generated when &optional or &aux
2240 variables have default values. (reported by Bruno Haible)
2241 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2242 have CLOS classes; this appears also to have fixed reported
2243 instability in stack exhaustion detection.
2244 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2245 MUFFLE-WARNING when no associated restart is present is now
2247 * optimization: rearranged the expansion of various defining macros
2248 so that each expands into only one top-level form in a
2249 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2251 * optimization: used a previously little-used slot in symbols to
2252 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2254 * fixed some bugs revealed by Paul Dietz' test suite:
2255 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2257 ** Fixed an optimization bug related to inheritance of initforms
2258 from local to shared slots.
2259 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2260 ** CAST optimizer forgot to flush argument derived type.
2261 ** print/read consistency on floats is now orders of magnitude
2262 more likely. (thanks also to Bruno Haible for a similar report
2264 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2265 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2266 for objects of type REAL. Make it so.
2267 ** SXHASH is specified (infelicitously) to respect similarity,
2268 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2269 it so. (thanks to Markus Ziegler)
2270 ** on the Alpha, the compiler succeeds in compiling functions
2271 returning a known number of arguments greater than 63.
2272 ** fixed handling of invalid NIL arguments in keyword position
2274 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2276 ** provided a readably-printable representation for RANDOM-STATE
2278 ** ensured that pathnames, where they have a namestring, always
2279 print using #P"..." syntax.
2281 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2282 * Support for the forthcoming 2.0 version of the NetBSD kernel
2283 running on x86 hardware has been added. (thanks to Perry
2284 E. Metzger most immediately, and others for their past work)
2285 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2286 current release version; SBCL's OpenBSD support had been broken
2287 since about the time of OpenBSD's switch to ELF binary format.)
2288 * [placeholder for DX summary]
2289 ** user code with &REST lists declared dynamic-extent, under high
2290 speed or space and low safety and debug optimization policy.
2291 * The manual has been converted to Texinfo format and the debugger
2292 chapter from the cmucl manual has been added.
2293 * A facility has been added to extract documentation strings from
2294 sbcl and store them as Texinfo-formatted snippets for inclusion in
2295 the manual (via Texinfo's @include directive)
2296 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2297 the test case to Dave Roberts)
2298 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2299 pointers. (reported by Sean Ross)
2300 * bug fix: PROFILE output is printed nicely even for large numerical
2301 values. (thanks to Zach Beane)
2302 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2303 greater than 32 handle EOF correctly.
2304 * bug fix: on X86 an immediate argument of the IMUL instruction is
2305 correctly printed in disassembly. (thanks to Lutz Euler)
2306 * bug fix: class slots in redefined classes preserve their old
2307 values. (thanks to Bruno Haible and Nikodemus Siivola)
2308 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2309 succeeds. (reported by Marco Baringer)
2310 * fixed some bugs revealed by Paul Dietz' test suite:
2311 ** READ-SEQUENCE now works on ECHO-STREAMs.
2312 ** RATIONALIZE works more according to its specification. (thanks
2315 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2316 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2317 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2318 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2319 should matter to you only if you rebind the printer control
2320 variables and then find you want different bindings in the
2321 debugger than in the ordinary execution of your program.)
2322 * The runtime build system has been tweaked to support building
2323 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2324 assembler and linker. (thanks to Nikodemus Siivola)
2325 * Unbound, undefined, undeclared variables now trigger full
2326 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2327 more often programmer error than deliberate exploitation of undefined
2329 * optimization: the hash algorithm for strings has changed to one
2330 that is less vulnerable to spurious collisions. (thanks to Juho
2332 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2333 needless bounds checking. (thanks to Juho Snellman)
2334 * optimization: implemented multiplication as a modular
2335 (UNSIGNED-BYTE 32) operation on the PPC backend.
2336 * fixed some bugs revealed by Paul Dietz' test suite:
2337 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2339 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2340 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2341 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2343 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2345 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2347 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2348 their output stream on EOF from read.
2349 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2350 have been read to end-of-file.
2351 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2353 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2354 description of determination of which consecutive characters
2356 ** Printing the "Space" character with escaping on now yields "#\\ ",
2357 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2358 ** Reading floating-point numbers with *READ-BASE* set to a number
2359 less than 10 works correctly.
2360 ** Reading floating-point numbers with *READ-BASE* set to a number
2361 more than 10 works correctly.
2362 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2363 the readtable currently in effect.
2365 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2366 * minor incompatible change: parsing of namestrings on a physical
2367 (Unix) host has changed; numbers after the final #\. in a
2368 namestring are no longer interpreted as a version field. This is
2369 intented to be largely invisible to the user, except that the
2370 meaning of the namestring "*.*.*" has changed: it now refers to a
2371 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2372 should usually be replaced by
2373 (make-pathname :name :wild :type :wild :version :wild)
2374 with the added benefit that this is more likely to be portable.
2375 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2376 option to OPEN now signals an error if the file being opened
2377 exists; this may have an impact on existing code.
2378 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2379 signals received, on the PowerPC platforms (both Linux and
2380 Darwin). (thanks to Pierre Mai for pointing out the location of
2382 * several fixes on OS X: The system now builds and runs cleanly on
2383 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2384 (thanks to Brian Mastenbrook)
2385 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2386 second-resolution: integer multiples of 1/3600 between -24 and 24.
2387 (thanks to Vincent Arkesteijn)
2388 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2389 type when called with 1 argument; PEEK-CHAR checked type of
2390 PEEK-TYPE only after having read first character from a
2391 stream. (reported by Peter Graves)
2392 * bug fix: the garbage collector now has much better locality
2393 behaviour, and in particular no longer treats all memory as being
2394 exhausted when many small objects point to each other in a deeply
2396 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2397 recognized as being TYPEP their class.
2398 * bug fix: the PUSHNEW documentation string has been corrected.
2399 (thanks to Vincent Arkesteijn)
2400 * bug fix: defaulting of the value for the last of an atypically
2401 large number of multiple values being bound was not being
2402 performed correctly on the Alpha or PPC platforms
2403 * optimization: implemented multiplication as a modular
2404 (UNSIGNED-BYTE 32) operation on the x86 backend.
2405 * optimization: SEARCH on simple-base-strings can now be open-coded.
2406 (see also contrib/compiler-extras.lisp for inspiration for
2407 teaching the compiler about the Boyer-Moore algorithm).
2408 * value, returned by MAX (and MIN) called with several EQUALP, but
2409 not EQL, arguments now does not depend on compiler settings.
2410 * fixed some bugs revealed by Paul Dietz' test suite:
2411 ** in stack analysis liveness information is propagated from
2412 non-local entry points.
2413 ** pathwise CAST removing failed when the CAST node did not start
2415 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2416 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2418 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2419 host is already defined.
2420 ** RENAME-FILE works on streams instead of signalling an internal
2422 ** PEEK-CHAR uses the current readtable when determining whether
2423 or not a character is whitespace.
2424 ** MERGE-PATHNAMES handles the case when the pathname does not
2425 specify a name while the default-pathname specifies a version.
2426 ** Pathnames now stand a better chance of respecting print/read
2428 ** Attempting to use standardized file system operators with a
2429 pathname with invalid :DIRECTORY components signals a
2431 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2433 ** DIRECTORY on logical pathnames is more correct.
2434 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2435 signal a TYPE-ERROR if their argument is not a stream.
2436 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2437 designator argument does not designate a stream.
2438 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2439 examining the synonym.
2440 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2442 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2443 element-type larger than ([UN]SIGNED-BYTE 32).
2445 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2446 * When built with the :SB-FUTEX feature, threaded builds now take
2447 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2448 for faster/more reliable mutex and condition variable support.
2449 * Incompatible change (but one you probably shouldn't have been using
2450 anyway): the interface and code for arbitrating between multiple
2451 threads in the same user session has been redesigned.
2452 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2453 error when called without an explicit environment argument.
2454 (thanks to Nikodemus Siivola)
2455 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2456 confused about their position. (thanks to Adam Warner and Gerd
2458 * bug fix: Pretty printing backquoted forms with unquotations in the
2459 argument list position of various code constructs such as LAMBDA
2460 now works correctly. (reported by Paul Dietz)
2461 * bug fix: Pretty printing unquotations no longer loses all
2462 stream position information.
2463 * optimization: performance of string output streams is now less
2464 poor for multiple small sequence writes.
2465 * optimization: performance of CSUBTYPEP in the presence of complex
2466 expressions involving CONS and NOT many times has been improved.
2467 (reported by Paul Dietz)
2468 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2470 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2471 streams for the REPL, for the debugger, and so on. (thanks to
2473 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2474 * fixed some bugs revealed by Paul Dietz' test suite:
2475 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2476 is ignored if it is too small, rather than propagating through
2477 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2478 ** extremely complex negations of CONS types were not being
2479 sufficiently canonicalized, leading to inconsistencies in
2481 ** VALUES tranformer lost derived type.
2483 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2484 * fixed a bootstrapping bug: the build process no longer assumes
2485 that the various BOOLE-related constants have the same value in
2486 host and target lisps. (noted by Paul Dietz' test suite on an
2487 SBCL binary built from CLISP)
2488 * The system can now be dynamically linked on the MIPS platform,
2489 which enables dynamic loading of foreign code from Lisp. (thanks
2490 to Ralf Baechle for discussions on the MIPS ABI)
2491 * The system now records debugging information for its own source
2492 files in a filesystem-position-independent manner, relative to
2493 the "SYS" logical host.
2494 * fixed a compiler bug: MV-LET convertion did not check references
2495 to the "max args" entry point. (reported by Brian Downing)
2496 * tweaked disassembly notes to be less confident about proclaiming
2497 some instruction as an LRA. (thanks to Brian Downing)
2498 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2499 now each have their own history, command character, and other
2500 characteristics. (thanks to David Lichteblau)
2501 * fixed some bugs revealed by Paul Dietz' test suite:
2502 ** compiler failure in compiling LOGAND expressions including a
2504 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2505 shift greater than 32.
2506 ** FLUSH-DEST did not mark blocks for type check regeneration.
2507 ** HANDLER-CASE failed to accept declarations in handler clauses
2508 in some circumstances.
2510 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2511 * New code in contrib/sb-introspect (still probably not entirely
2512 stable yet) provides some support for smart Lisp development
2513 environments like SLIME.
2514 * The conditions signalled for errors occurring when loading .fasl
2515 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2516 in a way which should help ASDF recover gracefully.
2517 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2518 stuff has been cleaned up. If you code contrib/ stuff, this might
2519 affect you, and you can look at contrib/README, contrib/STANDARDS,
2520 and/or the 0.8.4.27 diff to check.
2521 * In full calls the compiler now does not generate checks for declared
2522 argument types for all arguments.
2523 * various threading fixes
2524 ** and some experimental patches which didn't make it into
2525 the main tree for this release, but which are shipped in
2526 contrib/experimental-thread.patch as a possible fix for some
2527 failures (deadlock, spinning...) in GC-intensive multithreaded
2529 * fixed PPC build problem (source code incompatibility of different
2530 library versions): added offsetof() hackery which attempts to divine
2531 where glibc maintainers put uc_mcontext today
2532 * fixed bug 282: compiler does not trust type assertions while passing
2533 arguments to a full call.
2534 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2535 &OPTIONAL VALUES type parameter.
2536 * fix bug 214: algorithm for noting rejected templates is now more
2537 similar to that of template seletion. (also reported by rydis on
2539 * fixed bug 141b: printing backquoted information readably and prettily
2540 inserts a space where necessary.
2541 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2542 (CHARACTER 10) are now reported as errors, rather than propagated
2543 as unknown types. (reported by piso on #lisp)
2544 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2545 with values NIL and :ERROR. (thanks to Milan Zamazal)
2546 * fixed bug 191c: CLOS now does proper keyword argument checking as
2547 described in CLHS 7.6.5 and 7.6.5.1.
2548 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2549 longer raise an error; further, using a list as a for-as-arithmetic
2550 counter now raises a meaningful error.
2551 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2552 now understood by sequence creation functions such as MAKE-SEQUENCE
2554 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2555 a STRING-INPUT-STREAM.
2556 * compiler enhancement: SIGNUM is now better able to derive the type
2558 * type declarations inside WITH-SLOTS are checked. (reported by
2560 * fixed some bugs revealed by Paul Dietz' test suite:
2561 ** incorrect optimization of TRUNCATE for a positive first
2562 argument and negative second.
2563 ** compiler failure in let-convertion during flushing dead code.
2564 ** compiler failure while deriving type of TRUNCATE on an
2565 interval, containing 0.
2566 ** ASH of a negative bignum by a negative bignum count now returns
2568 ** intersection of CONS types now canonicalizes properly, fixing
2569 inconsistencies in SUBTYPEP.
2571 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2572 * incompatible change: The --disable-debugger command line
2573 option now clobbers the debugger at a more fundamental
2574 level, by redefining #'INVOKE-DEBUGGER instead of by
2575 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2576 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2577 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2578 enter the debugger REPL and then suffer recursive errors
2579 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2580 while under the new system, BREAK in --disable-debugger mode
2581 terminates the system just as an unhandled error would.
2582 * fixed compiler performance when processing loops with a step >1;
2583 * bug fix: DOCUMENTATION now retrieves generic function
2584 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2585 support has been systematized, and now supports the methods
2586 specified by ANSI, along with a default method and a method for
2587 slot documentation. (reported by Nathan Froyd)
2588 * bug fix: effective methods associated with a generic function are
2589 no longer cached over a change of that generic function's method
2590 combination. (reported by Andreas Fuchs)
2591 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2592 names a non-existent directory. (thanks to Andreas Fuchs)
2593 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2594 the PPC platform, lead to stack corruption; this has been fixed.
2595 (reported by Rainer Joswig)
2596 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2597 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2598 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2599 arithmetic in general, is now much more reliable on the PPC
2601 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2602 platform now returns the right answer.
2603 * optimization: restored some effective method precomputation in
2604 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2605 precomputation is now tunable.
2606 * optimization: compiler-internal data structure use has been
2607 reviewed, and changes have been made that should improve the
2608 performance of the compiler by about 20%.
2609 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2610 simple-base-strings and simple-bit-vectors is improved.
2611 * optimization: the optimization of 32-bit logical and arithmetic
2612 functions introduced in version 0.8.3 on the x86 has been
2613 implemented on the mips, ppc and sparc platforms; an
2614 implementation of the same facility, but for 64-bit arithmetic,
2615 has been added for the alpha.
2616 * microoptimization: the compiler is better able to make use of the
2617 x86 LEA instruction for multiplication by constants.
2618 * bug fix: in some situations compiler did not report usage of
2619 generic arithmetic in (SPEED 3) policy.
2620 * bug 145b fix: compiler used wrong type specifier while converting
2621 MEMBER-types to numeric.
2622 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2624 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2625 index argument now works properly on the Alpha platform.
2626 * bug fix: floating point exception treatment on the Alpha platform
2628 * bug fix: FILE-POSITION works much better on string input and
2629 output streams. (thanks to Nikodemus Siivola)
2630 * bug fix: many threading/garbage collection symptoms sorted.
2631 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2632 might be pseudo-atomic.
2633 * internal change: Stopping for GC is now done with signals not
2634 ptrace. GC is now done in whichever thread wanted it, instead of
2636 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2638 * bug fix: result form in DO is not contained in the implicit
2640 * incompatible change: ICR structure is changed; the value part of
2641 CONTINUATION is now called LVAR; corresponding functions are
2642 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2643 * added type deriver for ISQRT (thanks to Robert E. Brown).
2644 * bug fix: better support for loading from the command line when an
2645 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2647 * fixed some bugs revealed by Paul Dietz' test suite:
2648 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2649 ** ROUND and FROUND now give the right answer when given very
2650 small float arguments.
2651 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2653 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2654 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2655 ** LCM should return a non-negative integer.
2656 ** PARSE-INTEGER returned the index of a terminator instead of the
2657 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2658 ** PARSE-INTEGER returned an incorrect index being applied to a
2660 ** LCM with two arguments of 0 returns 0 rather than signalling
2662 ** unsigned addition of a 32-bit constant with the high bit set no
2663 longer causes an internal compiler error on the x86.
2664 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2665 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2666 with negative last argument.
2667 ** byte specifiers with zero size and position no longer cause
2668 an error during type derivation.
2669 ** bignum multiplication on the Alpha platform now returns the
2671 * porting: The system now builds on SuSE AMD64, although it still
2672 generates a 32-bit binary.
2673 * .fasl file incompatibility: The fasl file version number has
2674 been incremented (because of the changes to internal compiler
2675 data structures referred to above).
2677 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2678 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2679 more accurately, on the Darwin kernel running on PowerPC hardware.
2680 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2681 * Compiler code deletion notes now signal a condition of type
2682 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2683 an associated MUFFLE-WARNING restart.
2684 * The compiler now performs limited argument count validation of
2685 constant format strings in FORMAT, and where appropriate in ERROR,
2686 CERROR and WARN. (thanks to Gerd Moellmann)
2687 * New ASDF-INSTALL contrib can be used for automatic download and
2688 installation of third-party Lisp code from CCLAN or other sites
2690 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2691 which forces another thread to execute a function supplied by the
2693 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2694 now accept and act upon their :ELEMENT-TYPE keyword argument.
2695 (reported by Martin Atzmueller, Edi Weitz)
2696 * bug fix: FILE-POSITION now accepts position designators up to
2697 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2698 is the greater. (thanks to Patrik Nordebo)
2699 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2700 Kalle Olavi Niemitalo)
2701 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2702 printable. (reported by Eric Marsden)
2703 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2704 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2705 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2706 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2707 (thanks to Stig Sandoe)
2708 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2709 UNREAD-CHAR now correctly includes the unread character in the
2710 target sequence. (thanks to Gerd Moellmann)
2711 * bug fix in threaded builds: the system can now be suspended and
2712 resumed by shell job control with minimal disruption.
2713 * bug fixes in times and timezones >2038AD
2714 * better handling of "where is GNU make?" problem in build scripts
2715 (thanks to Nikodemus Siivola)
2716 * new optimization: inside a named function any reference to a
2717 function with the same name is considered to be a self-reference;
2718 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2719 optimization quality.
2720 * new optimization on x86: logical functions and + now have
2721 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2722 used when the result is truncated to 32 bits.
2723 * VALUES declaration is partially enabled.
2724 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2725 * fixed some bugs revealed by Paul Dietz' test suite:
2726 ** The system now obeys the constraint imposed by
2727 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2728 types form a lattice under type intersection.
2729 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2730 ** ASSOC now ignores NIL elements in an alist.
2731 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2732 and (1+ MOST-POSITIVE-FIXNUM) answers.
2733 ** The addition of a method with invalid qualifiers to a generic
2734 function does not cause an error to be signalled immediately;
2735 a warning is signalled, and the error is generated only on
2736 calling the generic function.
2737 * changed .fasl file version number, in part to add type codes for
2738 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2739 obscure ANSI requirements
2741 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2742 * fixed bug 148: failure to inline-expand a local function left
2743 garbage, confusing the compiler.
2744 * fixed bugs 3cd: structure slot readers perform type check if the
2745 slot can have an invalid value (i.e. it is either not initialized
2746 or can be written with a less specific slot writer).
2747 * bug fix: the compiler now traps array references to elements off
2748 the end of an array; previously, the bounds checking in some
2749 circumstances could go off-by-one.
2750 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2752 * type declarations for array element types now obey the description
2753 on the CLHS page "Declaration TYPE", as per discussions on
2754 sbcl-help around 2003-05-08. This means that a declaration
2755 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2756 declaration, all references to BAR will be asserted or assumed
2757 (with THE, so dependent on compiler policy) to involve objects of
2758 type FOO. Note that no such declaration is implied in
2759 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2760 * declared types of functions from the "Conditions"
2761 chapter. (reported by Paul Dietz)
2762 * bug fix: CERROR accepts a function as its first argument.
2763 * bug fix: NTH an NTHCDR accept a bignum as index
2764 arguments. (reported by Adam Warner)
2765 * optimization: character compare routines now optimize comparing
2766 against a constant character. (reported by Gilbert Baumann)
2767 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2768 argument now works properly on the MIPS platform.
2769 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2770 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2771 declared to have a functional type, does not cause an error
2773 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2774 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2775 * fixed bug: if last continuation of a deleted block has a
2776 destination, this destination should be deleted too. (reported by
2778 * fixed a bug in the bootstrap process: the host compiler's values
2779 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2780 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2781 test case from Patrik Nordebo)
2782 * improved the ability of the disassembler on the PPC platform to
2783 provide helpful disassembly notes.
2784 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2785 the class in more cases than previously.
2786 * bug fix: FILE-POSITION now understands :START and :END for
2787 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2788 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2789 8) data. (thanks to David Lichteblau)
2790 * bug fix: it is possible to add a method to a generic function
2791 without lambda list.
2792 * bug fix: reader failed to signal END-OF-FILE inside an
2793 object representation. (reported by Nikodemus Siivola)
2794 * fixed some bugs revealed by Paul Dietz' test suite:
2795 ** LAST and [N]BUTLAST should accept a bignum.
2796 ** condition slot accessors are methods.
2797 ** (VECTOR NIL) is a subtype of STRING.
2799 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2800 * minor incompatible change: some nonsensical specialized lambda
2801 lists (used in DEFMETHOD) which were previously ignored now signal
2803 * minor incompatible change: the system is now aware of the types of
2804 variables in the COMMON-LISP package, and will signal errors for
2805 most violations of these type constraints (where previously they
2806 were silently accepted).
2807 * minor incompatible change: COMPILE-FILE now uses the freedom
2808 afforded (ANSI 3.2.2.3) to use derived function types for
2809 functions defined in the same file. This also permits the system
2810 to warn on static type mismatches and function
2811 redefinition. (Currently it does not work with high DEBUG level.)
2812 * minor incompatible change: VALUES declaration is disabled.
2813 * When issuing notes, the compiler now signals a condition of type
2814 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2815 restart for use in user handlers. It is expected that the
2816 COMPILER-NOTE condition will eventually become a condition
2817 supertype to a hierarchy of note types, which will then be
2818 handleable in a similar fashion. However, at the moment, no such
2819 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2820 but it's an implementation detail, not a classification for the
2822 * Changes in type checking closed the following bugs:
2823 ** type checking of unused values (192b, 194d, 203);
2824 ** template selection based on unsafe type assertions (192c, 236);
2825 ** type checking in branches (194bc).
2826 * A short form of VALUES type specifier has ANSI meaning (it has
2827 increased the number of situations when SBCL cannot perform type
2829 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2830 keywords or constants is permissible.
2831 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2832 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2833 classes. (thanks to Antonio Martinez)
2834 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2835 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2836 argument to a pretty-print stream. Instead, it leaves any such
2837 operations to DESCRIBE-OBJECT methods.
2838 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2840 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2841 lambda lists. (thanks to David Lichteblau)
2842 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2843 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2844 (thanks to Gerd Moellmann)
2845 * fixed bug 246: increased compilation speed of long
2846 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2848 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2850 * DEFCONSTANT now throws a condition of type
2851 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2852 constant to a non-EQL value; CONTINUE and ABORT restarts
2853 respectively change and preserve the value.
2854 * fixed bug 63: The code walker, part of the implementation of CLOS,
2855 is now better at handling symbol macros.
2856 * bug fix: There is no longer an internal implementation type named
2857 CL:LENGTH. (reported by Raymond Toy)
2858 * bug fix: In macro-like defining macros/special operators the
2859 implicit block does not enclose the lambda list.
2860 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2861 enclosed in parentheses) are not suitable as type specifiers, and
2862 their use properly signals an error now.
2863 * bug fix: An argument count mismatch for a type specifier in code
2864 being compiled no longer causes an unhandled error at compile
2865 time, but signals a compile-time warning.
2866 * fixed simple vector readable printing
2867 * bug fix: DESCRIBE takes more care over whether the class
2868 precedence list slot of a class is bound before accessing it.
2869 (reported by Markus Krummenacker)
2870 * bug fix: FORMATTER can successfully compile pretty-printer format
2871 strings which use variants of the ~* directive inside.
2872 * bug fix: SEARCH now applies its TEST predicate to the elements of
2873 the arguments in the correct order. (thanks to Wolfhard Buss)
2874 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2875 in type inference. (thanks to Robert E. Brown)
2876 * bug fix: Reading in symbols with an explicit package name of ""
2877 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2878 * fixed some bugs revealed by Paul Dietz' test suite:
2879 ** NIL is now allowed as a structure slot name.
2880 ** Arbitrary numbers, not just REALs, are allowed in certain
2881 circumstances in LOOP for-as-arithmetic clauses.
2882 ** Multiple class redefinitions before slot access no longer
2883 causes a type error.
2884 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2885 association between the name and a class.
2886 ** Generic functions with non-standard method-combination and over
2887 five methods all of which return constants no longer return NIL
2888 after the first few invocations. (thanks to Gerd Moellmann)
2889 ** CALL-NEXT-METHOD with no arguments now passes the original
2890 values of the arguments, even in the presence of assignment.
2891 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2893 ** DISASSEMBLE works with closures and funcallable instances.
2894 ** ADD-METHOD now returns the generic function, not the new method.
2895 ** FIND-METHOD signals an error if the lengths of the specializers
2896 is incompatible with the generic function, even if the ERRORP
2898 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2899 which its argument is a member.
2900 ** DEFCLASS only redefines the class named by its class-name
2901 argument if that name is the proper name of the class;
2902 otherwise, it creates a new class.
2903 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2904 of the UNBOUND-SLOT condition to the name of the slot.
2905 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2906 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2907 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2908 SLOT-MAKUNBOUND in the specified fashion.
2910 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2911 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2912 cross-compilation host. As a consequence, we can now bootstrap our
2913 way up to SBCL starting with a bare gcc toolchain and human-readable
2914 source code (first the source to CLISP, then the source to SBCL).
2915 * A contributed module containing a partial implementation of the
2916 simple-streams interface has been included. (thanks to Rudi
2918 * A contributed module implementing the RFC1321 Message Digest
2919 Algorithm, known as MD5, has been included.
2920 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2921 no longer has any effect, as the code controlled by this feature
2922 has been deleted. (As far as we know, no-one has ever built using
2923 this feature, and its semantics were confused in any case).
2924 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2925 work on conditions (as required by the ANSI specification),
2926 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2927 expected behaviour on conditions. Users should note, however,
2928 that such behaviour is not required by the ANSI specification,
2929 and so use of this behaviour may render their code unportable.
2930 * fixed some bugs revealed by Paul Dietz' test suite:
2931 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2933 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2934 ** SXHASH on condition objects no longer returns NIL.
2935 ** :ALLOCATION :CLASS slots are better treated; their values are
2936 updated on class redefinition, and initforms inherited from
2937 superclasses are applied.
2938 ** REMOVE-METHOD returns its generic function argument even when
2939 no method was removed.
2940 ** SHARED-INITIALIZE now initializes the values of the requested
2941 slots, including those with :ALLOCATION :CLASS.
2942 ** ALLOCATE-INSTANCE now works on structure classes defined via
2943 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2945 ** SLOT-EXISTS-P now works on conditions, as well as structures
2947 ** MAKE-LOAD-FORM now has the required methods on
2948 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2949 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2950 keyword, and now implements the SLOT-NAMES argument.
2951 ** methods with &OPTIONAL arguments no longer allow too many
2952 arguments to be passed in the call without error.
2953 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2954 option is consistent with the required arguments of the generic
2955 function lambda list.
2956 * bug fix: REQUIRE accepts a string designator. (Thanks to
2958 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2959 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2960 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2961 * bug fix: The compiler now checks for duplicated variables in macro
2963 * bug fix: SETQ on globals returns the correct value.
2964 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2965 not a COMPILER-ERROR (followed by some other strange error on
2966 choosing the CONTINUE restart).
2967 * bug fix: make.sh and friends are now more consistent in the way that
2968 they look for GNU "make".
2970 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2971 * experimental native threads support (on x86 Linux >=2.4 only).
2972 This is not compiled in by default: you need to add :SB-THREAD to
2973 the target features. See the "Beyond ANSI" chapter of the manual
2975 * fix for longstanding nonANSIism: The old distinction between
2976 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2977 The return value from CL:FIND-CLASS is now a CLOS class, and
2978 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2979 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2980 * An interface to the MetaObject Protocol, as described in Kiczales,
2981 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2982 MIT Press, 1991, is available from the SB-MOP package.
2983 * incompatible change: the SB-PCL package should now be considered
2984 a private implementation detail, and no longer a semi-private MOP
2986 * minor incompatible change: due to rearrangement for threads, the
2987 control stack and binding stack are now allocated at arbitrary
2988 addresses instead of being hardcoded per-port. Users affected by
2989 this probably have to be doing advanced things with shared
2990 libraries, and will know who they are.
2991 * minor incompatible change: Previously, all --eval forms used were
2992 processed with READ before any of them were processed with EVAL.
2993 Now each --eval form is processed with both READ and EVAL before
2994 the next --eval form is processed. (Thus package operations like
2995 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2996 work as the user might reasonably expect.)
2997 * minor incompatible change: *STANDARD-INPUT* is now only an
2998 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3000 * minor incompatible change: Y-OR-N-P is now character-oriented,
3001 not line oriented. Also, YES-OR-NO-P now works without errors.
3002 (thanks to Antonio Martinez)
3003 * sb-aclrepl module improvements: an integrated inspector, added
3004 repl features, and a bug fix to :trace command.
3005 * Known functions, which cannot be open coded by the backend, are
3006 considered to be able to check types of their arguments. (fixing
3007 a bug report by Nathan J. Froyd)
3008 * fixed a bug in computing method discriminating functions: It is
3009 now possible to define methods specialized on classes which have
3010 forward-referenced superclasses. (thanks to Gerd Moellmann)
3011 * fixed evaluation order in optional entries (reported by Gilbert
3013 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3014 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3015 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3016 required-by-AMOP NAME argument, as well as CLASS and
3017 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3018 * fixed bug 20: DEFMETHOD can define methods using names that are
3019 not the proper names of classes to designate class specializers.
3020 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3021 Unix file descriptors, instead of blowing up. (thanks to Antonio
3023 * Garbage collection refactoring: user-visible change is that a
3024 call to the GC function during WITHOUT-GCING will not do garbage
3025 collection until the end of the WITHOUT-GCING. If you were doing
3026 this you were probably losing anyway.
3027 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3028 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3029 TYPEP the latter but not the former.
3030 * The compiler issues a full WARNING for calls to undefined functions
3031 with names from the CL package.
3032 * MAP-INTO for a vector destination is open coded. (reported by
3033 Brian Downing on c.l.l)
3034 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3035 documentation string.
3036 * fixed some bugs revealed by Paul Dietz' test suite:
3037 ** COPY-ALIST now signals an error if its argument is a dotted
3039 ** Condition slots are now accessed more correctly in the presence
3040 of multiple initargs for a given slot.
3041 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3042 exclude restarts of the same name associated with a different
3044 ** DEFCLASS of forward-referenced classes with another
3045 forward-referenced class in the superclasses list no longer
3047 ** Condition slots are now initialized once each, not multiple
3048 times. (thanks to Gerd Moellmann)
3049 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3051 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3052 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3053 arguments contain duplicated elements.
3054 ** RESTART-CASE understands local macros.
3055 ** RESTART-CASE associates exactly its own restarts with a condition.
3056 ** ENDP in safe mode checks its argument to be of type LIST.
3057 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3058 in question is unbound.
3059 ** Optimized MAKE-INSTANCE functions no longer cause internal
3060 assertion failures in the presence of duplicate initargs.
3061 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3062 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3063 propagated even in the ctor optimized implementation of
3065 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3067 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3068 * a better implementation of SXHASH on (simple) bit vectors,
3069 measured both in execution speed and in distribution of results
3070 over the positive fixnums, has been installed. Likewise, a better
3071 implementation of EQUAL for simple bit vectors is now available.
3072 * fixed CEILING optimization for a divisor of form 2^k.
3073 * fixed bug 240 (emitting extra style warnings "using the lexical
3074 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3075 by Antonio Martinez)
3076 * fixed SXHASH, giving different results for NIL depending on type
3077 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3078 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3079 variable. (found by Rolf Wester)
3080 * SBCL does not ignore type declarations for special
3081 variables. (reported by rif on c.l.l 2003-03-05)
3082 * some bug fixes in contrib/sb-aclrepl/
3083 * fixed some bugs revealed by Paul Dietz' test suite:
3084 ** a bug in the CONS type specifier, whereby the CAR and CDR
3085 types got intertwined, has been fixed;
3086 ** the type system is now able to reason about the interaction
3087 between INTEGER and RATIO types more completely;
3088 ** APPEND, [N]REVERSE and NRECONC check that those their
3089 arguments, which must be proper lists, are really so;
3090 ** An array specialized to be unable to hold elements has been
3091 implemented, as required -- yes, really -- by ANSI;
3092 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3093 on malformed property lists;
3095 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3096 * incompatible packaging change: in line with Unix convention,
3097 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3098 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3099 * REQUIRE and PROVIDE are now optionally capable of doing something
3100 useful. See the documentation string for REQUIRE.
3101 * infrastructure for a managed SBCL contrib system: contributed
3102 modules in this release include:
3103 ** the ASDF system definition facility;
3104 ** an interface to the BSD Sockets API;
3105 ** an ACL-like convenience interface to the repl;
3106 (thanks to Kevin Rosenberg)
3107 ** an implementation of ROTATE-BYTE, with efficient implementation
3109 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3110 gives a closer approximation to the right answer than previously.
3111 (thanks to Raymond Toy)
3112 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3113 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3114 optional environment arguments, as required by ANSI.
3115 * fixed bugs in other functions taking environment objects, allowing
3116 calls with an explicit NIL environment argument to be compiled
3118 * fixed bug 228: primary return values from
3119 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3120 COMPILE or FUNCTION.
3121 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3122 :TYPE will no longer signal errors on innocuous objects.
3123 * fixed bug 231b: SETQ is better at respecting type declarations in
3124 the lexical environment.
3125 * fixed a bug in DEFCLASS: classes named by symbols with no or
3126 unprintable packages can now be defined.
3127 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3128 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3129 * fixed compiler failure related to checking types of functions.
3130 (reported by Robert E. Brown)
3131 * the compiler is now much more consistent in its error-checking
3132 treatment of bounding index arguments to sequence functions: in
3133 (SAFETY 3) code, errors will be signalled in almost all cases if
3134 invalid sequence bounding indices are passed to functions defined
3135 by ANSI to operate on sequences.
3136 * fixed a bug in the build procedure: documentation of SBCL-specific
3137 packages is now preserved and available in the final Lisp image.
3138 * lifted FDEFINITION lookup out of loops in the implementation of
3139 many list operations. (thanks to Robert E. Brown)
3140 * fixed a bug in the reader: the #n# reader macro now works for
3141 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3142 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3143 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3144 E. Brown and Tony Martinez respectively)
3145 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3146 SLOT-BOUNDP: in particular, it is now able to optimize them much
3147 better, and is now not vulnerable to having packages renamed.
3148 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3149 always called when a slot is not present in an instance. (thanks
3151 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3152 longer signal an error on unreading a character following EOF on
3153 the previous constituent stream. (thanks to Tony Martinez)
3154 * fixed some bugs revealed by Paul Dietz' test suite:
3155 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3156 not just nonnegative fixnums;
3157 ** the logical bit-array operators such as BIT-AND now accept an
3158 explicit NIL for their "opt-arg" argument (to indicate a
3159 freshly-consed result bit-array);
3160 ** ELT now signals an error on an invalid sequence index in safe
3162 ** the type system is now cleverer about negations of numeric
3163 types, and consequently understands the BIGNUM and RATIO types
3165 ** the type system is now cleverer about the interaction between
3166 INTEGER and RATIO types: while bugs still remain, many more
3167 cases are accurately computed;
3168 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3169 if it is in the last clause;
3170 ** CONSTANTLY now correctly returns a side-effect-free function in
3172 ** DECLARE is no longer treated as a special-operator; in
3173 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3174 * incremented fasl file version number due to the change in the
3175 DEFSTRUCT-SLOT-DESCRIPTION structure.
3177 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3178 * minor incompatible change: code processed by the "interpreter" or
3179 EVAL now has a compilation optimization policy of (DEBUG 2)
3180 (changed from (DEBUG 1)) to improve debuggability of interactive
3181 development, and to allow the use of the debug RETURN command in
3183 * an experimental implementation of the RETURN command for the
3184 debugger has been included. (thanks to Frederik Kuivinen)
3185 * fixed bug 62: constraints were not propagated into a loop.
3186 * fixed bug in embedded calls of SORT (reported and investigated by
3188 * fixed some bugs revealed by Paul Dietz' test suite:
3189 ** printing and reading of arrays with some dimensions having
3190 length 0 (thanks to Gerd Moellmann);
3191 ** BOA constructor with &AUX argument without a default value does
3192 not cause a type error;
3193 ** CONSTANTP now returns true for all self-evaluating objects.
3195 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3196 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3197 accessors that are related by inheritance, as specified in the
3198 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3199 Valtteri Vuorikoski)
3200 * The compiler is now able to inline functions that were defined in
3201 a complex lexical environment (e.g. inside a MACROLET).
3202 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3204 * Support for the upcoming FreeBSD-5.0 release has been included.
3205 (thanks to Dag-Erling Smorgrav)
3206 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3207 effect when it is not in a toplevel context.
3208 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3209 stand a better chance of being correct. (thanks to Gerd
3211 * fixed bug in COERCE, which now signals an error on coercing a
3212 rational to a bounded real type which excludes the expected
3214 * The compiler is now able to derive types more accurately from the
3215 COERCE and COMPILE functions.
3216 * fixed bug 223: functional binding is considered to be constant
3217 only for symbols in the CL package.
3218 * fixed bug 231: SETQ did not check the type of a variable being set
3219 (reported by Robert E. Brown)
3220 * A new optimization for MAKE-INSTANCE has been included, fixing
3221 various bugs (including relating to :ALLOCATION :CLASS slots and
3222 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3224 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3225 ** As required by ANSI, LOOP now disallows anonymous collection
3226 clauses such as COLLECT I in conjunction with aggregate boolean
3227 clauses such as THEREIS (= I 1);
3228 ** LOOP now signals an error when any variable is reused in the
3229 same loop (including the potentially useful construct analogous
3230 to WITH A = 1 WITH A = (1+ A);
3231 ** IT is only a special loop symbol within the first clause of a
3232 conditional loop clause;
3233 ** LOOP with a typed iteration variable over a hashtable now
3234 signals a type error iff it should.
3235 * fixed some other bugs revealed by Paul Dietz' test suite:
3236 ** FILE-STREAM now names the class previously known as FD-STREAM;
3237 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3238 argument) no longer signals an error;
3239 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3240 of the slot symbol, rather than using the current package
3241 ((:CONC-NAME "") continues to intern the slot's name in the
3243 * incremented fasl file version number, because of the incompatible
3244 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3245 of the new implementation of DEFINE-COMPILER-MACRO.
3247 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3248 * Support for building SBCL for MIPS platforms running in
3249 little-endian mode has now been checked in, and basic
3250 functionality on said platforms verified.
3251 * minor incompatible change: PCL now records the pathname of a file
3252 in which methods and the like are defined, rather than its
3254 * minor incompatible change: TRUENAME now considers the truename of
3255 a file naming a directory to be the pathname with :DIRECTORY
3256 component indicating that directory.
3257 * minor incompatible change: a NAMED clause in the extended form of
3258 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3259 reason for the previous behaviour is unclear.
3260 * more systematization and improvement of CLOS and MOP conformance
3261 in PCL (thanks to Gerd Moellman and Pierre Mai):
3262 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3264 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3265 primary methods with no specializers;
3266 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3268 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3269 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3270 FORWARD-REFERENCED-CLASSes; error reporting on
3271 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3273 ** SXHASH on CLOS instances now uses a slot internal to the
3274 instance to return different numbers on distinct instances,
3275 while preserving the same return value through invocations of
3277 ** DEFMETHOD signals errors when methods with longer incongruent
3278 lambda lists are added to generic functions;
3279 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3280 CLASS, as specified in AMOP;
3281 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3282 on the order returned by the primary method for classes of
3283 class STANDARD-CLASS;
3284 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3285 * fixed some bugs shown by Paul Dietz' test suite:
3286 ** DOLIST puts its body in TAGBODY;
3287 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3289 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3290 value producing form;
3291 ** if more variables are given to PROGV than values, extra
3292 variables are bound and made to have no value;
3293 ** NSUBSTITUTE on list arguments gets the right answer with
3295 ** ELT signals an error of type TYPE-ERROR when the index argument
3296 is not a valid sequence index;
3297 ** LOOP signals (at macroexpansion time) an error of type
3298 PROGRAM-ERROR when duplicate variable names are found;
3299 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3300 ** FUNCALL of special-operators now cause an error of type
3302 ** PSETQ now works as required in the presence of side-effecting
3303 symbol-macro places;
3304 ** NCONC accepts any object as its last argument;
3305 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3307 ** loop-for-as-package does not require a package to be explicitely
3309 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3311 * fixed bug 166: compiler preserves "there is a way to go"
3312 invariant when deleting code.
3313 * fixed bug 172: macro lambda lists with required arguments after
3314 &REST arguments now cause an error to be signalled. (thanks to
3316 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3317 bug. (thanks to Matthew Danish)
3318 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3320 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3321 arguments are lexically rebound. (thanks to Gerd Moellmann and
3323 * fixed bug 194: error messages are now more informative when there
3324 is no primary method applicable in a call to a generic
3325 function. (thanks to Gerd Moellmann)
3326 * fixed bug in command line argument checking (thanks to Julian
3328 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3329 (thanks to Matthew Danish)
3330 * incremented fasl file version number, because of the
3331 SXHASH-related changes in the layout of CLOS data structures
3333 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3334 * minor incompatible change: The runtime (the Unix executable named
3335 "sbcl") is now much pickier about the .core files it will load.
3336 Essentially it now requires .core files to descend from the same
3337 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3338 as the runtime does. (The intent is to prevent the crashes which
3339 can occur, and which can even be reported as mysterious failures,
3340 when people patch the sources or change the build parameters
3341 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3342 sbcl and .core files.)
3343 * fixed bug: VALUES-LIST is no longer optimized away.
3344 * fixed bug 142: The FFI conversion of C string values to Lisp
3345 string values no longer conses excessively. (thanks to Nathan
3346 Froyd porting Raymond Toy's fix to CMU CL)
3347 * began to systematize and improve MOP conformance in PCL (thanks to
3348 Nathan Froyd, Gerd Moellman and Pierre Mai):
3349 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3351 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3352 ** FINALIZE-INHERITANCE is now called on class finalization;
3353 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3354 argument precedence order.
3355 * fixed bug 202: The compiler no longer fails on functions whose
3356 derived types contradict their declared type.
3357 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3358 so it can be non-toplevel.
3359 * The fasl file version number has changed (because of the new
3360 implementation of DEFMACRO).
3361 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3362 safe code, that any length requirement by their type-specifier
3363 argument is valid. The exceptions to this are described in bug
3365 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3366 functions now check the types of their inputs as required by ANSI.
3367 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3368 introduced symbol is DECLAREd to be SPECIAL.
3369 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3370 * fixed bug: :COUNT argument to sequence functions may be negative
3371 * fixed bug: body of DO-SYMBOLS may contain declarations
3372 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3373 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3374 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3375 symbol macro only once
3376 * fixed printing of call frame when argument list is unavailable
3377 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3378 * compiler no longer signals WARNING on unknown keyword
3381 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3382 * A beta-quality port to the mips architecture running Linux,
3383 based on the old CMUCL backend, has been made. It has been tested
3384 on a big-endian kernel, and works sufficiently well to be able to
3385 rebuild itself; it has not been tested in little-endian mode.
3386 * fixed an inconsistency between gencgc.c and purify.c which made
3387 dumping/loading .core files unreliable
3388 * fixed bug 120a: The compiler now deals correctly with IFs where
3389 the consequent is the same as the alternative, instead of
3390 misderiving the return type. (thanks to Alexey Dejneka)
3391 * fixed bug 113: Logical pathnames are now dumpable (the logical
3392 host is resolved at load-time, throwing an error if it is not
3394 * fixed bug 174: FORMAT's error message is slightly clearer when a
3395 non-printing character is used in a format directive.
3396 * fixed several bugs in compiler checking of type declarations, i.e.
3397 violations of the Python "declarations are assertions" principle
3398 (thanks to Alexey Dejneka)
3399 * fixed several bugs in PCL's error checking (thanks to Gerd
3401 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3403 * fixed bug in compilation of functions as first class values
3404 (thanks to Antonio Martinez-Shotton)
3405 * The compiler's handling TYPE-ERRORs which it can prove will
3406 inevitably happen at runtime has been cleaned up and corrected
3407 in several ways. (thanks to Alexey Dejneka)
3408 * improved argument type checking for various basic arithmetic
3409 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3410 lavished on them in the past that they can be compiled in many
3411 ways in different special cases
3412 * fixed bug 181: compiler checks validity of user supplied type
3414 * cleaned up code flushing in optimization: Function calls which
3415 should signal errors for safety purposes (e.g. which ANSI says
3416 should signal errors when their arguments are of incorrect type)
3417 are no longer optimized away.
3418 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3419 * incremented fasl file version number, because changes in the
3420 implementation of sequence functions like COERCE caused
3421 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3422 in old inline expansions) to become undefined. (Actually these
3423 changes were later undone, so we might very well be binary
3424 compatible with 0.7.7 after all, but leaving the version number
3425 incremented seemed like the simplest and most conservative
3428 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3429 * An alpha-quality port to the parisc architecture running Linux,
3430 based on the old CMUCL backend, has been made. This, even more so
3431 than the other backends, should be considered still a work in
3432 progress; known problems include that the Linux kernel in 64-bit
3433 mode does not propagate the correct sigcontext structure to
3434 userspace, and consequently SBCL on a parisc64 kernel will not
3436 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3437 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3438 suppressed.) Also now that the compiler is looking at declarations
3439 in the environment, it checks optimization declarations as well,
3440 and suppresses inlining when (> DEBUG SPEED).
3441 * More fixes have been made to treatment of floating point exception
3442 treatment and other Unix signals. In particular, floating point
3443 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3444 * The detection and handling of control stack exhaustion (infinite
3445 or very deeply nested recursion) has changed. Stack exhaustion
3446 detection is now done by write-protecting pages at the OS level
3447 and applies at all optimization settings; when found, a
3448 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3449 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3450 be caught using IGNORE-ERRORS.
3451 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3452 correctly when transforming recursive LABELS functions to LETs.
3453 (thanks to Alexey Dejneka)
3454 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3455 that are names of constants or global variables.
3456 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3457 alien routines with docstrings.
3458 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3459 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3461 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3462 (thanks to Eric Marsden)
3463 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3464 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3465 * Bug fix: The SPARC backend can now compile functions involving
3466 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3467 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3468 to a functions expecting a PATHNAME-DESIGNATOR.
3469 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3470 lambda lists. (thanks to Alexey Dejneka)
3471 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3472 functions so that EXPORTing the name of the class doesn't cause
3473 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3474 to Antonio Martinez for reporting this)
3475 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3476 constant-folded without failing an assertion. (thanks to Einar
3477 Floystad Dorum for reporting this)
3478 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3479 types are now dealt with more correctly by the compiler.
3480 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3481 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3482 behaviour analogous to RENAME-FILE. This puts its behaviour more
3483 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3485 * The fasl file version number has changed again. (because of the
3486 bug fix involving the names of PCL MAKE-INSTANCE functions)
3488 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3489 * bug fix: Floating point exceptions are treated much more
3490 consistently on the x86/Linux and PPC/Linux platforms.
3491 * Array initialization with :INITIAL-ELEMENT is now much faster for
3492 cases when the compiler cannot open code the array creation, but
3493 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3494 array accesses have also seen a speed increase.
3495 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3496 specified. (This was at the root of some bad interactions between
3497 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3499 * bug fix: Internal error arguments for undefined functions are now
3500 computed correctly on the PPC/Linux platform.
3501 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3502 Raymond Toy's patch for CMU CL)
3503 * Support for the Solaris 9 operating environment has been included
3504 (thanks to Daniel Merritt)
3505 * A very ugly but hopefully complete draft of the missing FFI chapter
3506 of the manual has been created by reformatting the corresponding
3507 CMU CL manual chapter into (currently very ugly and incoherent)
3508 DocBook and bringing it up to date for SBCL behavior. Thus, the
3509 manual is now essentially complete, at least by my extreme
3510 once-and-only-once standards, whereby it's acceptable to refer to
3511 the doc strings of SB-EXT functions as the primary documentation.
3512 * The fasl file version number has changed again, due to cleanup of
3513 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3514 is no longer a static symbol.)
3516 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3517 * SBCL now builds with OpenMCL (version 0.12) as the
3518 cross-compilation host; also, more progress has been made toward
3519 bootstrapping under CLISP.
3520 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3522 * bug 158 fixed: The compiler can now deal with integer loop
3523 increments different from 1; fixing this turned out also to fix
3525 * bug 169 fixed: no more bogus warnings about using lexical bindings
3526 despite the presence of perfectly good SPECIAL declarations (thanks
3527 to David Lichteblau)
3528 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3529 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3530 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3532 * bug 180 fixed: Method combination specifications no longer ignore
3533 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3534 * bug fix: Structure type predicate functions now check their argument
3535 count as they should.
3536 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3537 correctly. (thanks to Pierre Mai)
3538 * minor incompatible change: The --noprogrammer option is deprecated
3539 in favor of the new --disable-debugger option, which is very similar.
3540 (The major difference is that it takes effect at a slightly different
3541 time at startup, causing handling of errors in --sysinit and
3542 --userinit files will be affected differently.) The
3543 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3544 been added to allow this functionality to be controlled from ordinary
3545 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3546 maintainers, who might want to run non-interactive scripts to
3547 build SBCL cores which will later be used interactively.)
3548 * minor incompatible change: The LOAD function no longer, when given
3549 a wild pathname to load, loads all files matching that pathname.
3550 Instead, an error of type FILE-ERROR is signalled.
3552 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3553 * bug 147 fixed: The compiler preserves its block link/count
3554 invariants more correctly now so that it doesn't crash. (thanks
3556 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3558 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3559 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3560 no-value-for-this-character as NIL (as specified by ANSI).
3561 * HOST-NAMESTRING on physical pathnames now returns a string that is
3562 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3563 (thanks to Christophe Rhodes)
3564 * The Alpha port handles icache flushing more correctly. (thanks to
3566 * More progress has been made toward bootstrapping under CLISP. (thanks
3567 to Christophe Rhodes)
3568 * The fasl file format has changed again, because dynamic loading
3569 on OpenBSD (which has non-ELF object files) motivated some cleanups
3570 in the way that foreign symbols are transformed and passed around.
3571 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3572 is no longer treated as whitespace by the reader, but instead as
3573 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3574 |A
\x7fB|, instead of A as it used to.
3576 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3577 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3578 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3579 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3580 this as of 0.7.1.45, but was left out of the previous news section
3581 (thanks to Dan Barlow)
3582 * SBCL now runs on the Solaris operating system on SPARC architectures
3583 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3584 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3585 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3586 * SPARC backend cleanups, allowing builds of cores optimized for V8
3587 and V9 SPARCS, and also emission of code targeted to a particular
3588 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3590 * SBCL is closer to bootstrapping under CLISP, thanks to various
3591 fixes by Christophe Rhodes.
3592 * The fasl file format has changed again, to allow the compiler's
3593 INFO database to support symbol macros.
3594 * The user manual (in doc/) is formatted into HTML more nicely.
3595 (thanks to coreythomas)
3596 * The system is smarter about SUBTYPEP relationships, especially
3597 those involving NOT types (including types such as ATOM which are
3598 represented internally using NOT types). Thus SUBTYPEP is less
3599 likely to return (VALUES NIL NIL) in general, and in particular
3600 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3602 * The fasl file format has changed again, because the internal
3603 representation of types now includes a new slot to support the new
3604 SUBTYPEP-of-NOT-types logic.
3605 * (not a change in the main branch of SBCL, but a related prototype
3606 which can hopefully be merged into the main branch of SBCL in the
3607 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3608 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3609 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3610 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3611 dispatch macro character combinations. (thanks to Alexey Dejneka)
3612 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3613 that superficially look like logical namestrings correctly.
3614 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3616 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3617 default supplied-p) for &optional and &key arguments. (thanks to
3620 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3621 * incompatible change: The compiler is now less aggressive about
3622 tail call optimization, doing it only when (> SPACE DEBUG) or
3623 (> SPEED DEBUG). (This is an incompatible change because there are
3624 programs which relied on the old CMU-CL-style behavior to optimize
3625 away their unbounded recursion which will now die of stack overflow.)
3626 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3627 for non-GENCGC systems has been increased to 20M (since that
3628 seems much closer to the likely performance optimum for modern
3629 systems than the old 4M value was)
3630 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3631 and *DEBUG-PRINT-LEVEL*
3632 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3633 of CMU CL's support for SPARC, and various endianness and other
3634 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3635 * new syntactic sugar for the Unix command line: --load foo.bar is now
3636 an alternate notation for --eval '(load "foo.bar")'.
3638 ** The system now detects stack overflow and handles it gracefully,
3639 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3640 optimization settings. (This is a good thing in general, and
3641 its introduction in this version should be particularly timely
3642 for anyone whose code fails because of suppression of tail
3644 ** The system now hunts for the C variable "environ" in a more
3645 devious way, to avoid segfaults when the C library version differs
3646 between compile time and run time. (thanks to Christophe Rhodes)
3647 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3648 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3649 ** The compiler no longer issues bogus style warnings for undefined
3650 classes in the same source file as the DEFCLASSes which defined
3651 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3653 ** fixes in CONDITION class precedence list for undefined function
3654 errors (thanks to Alexei Dejneka)
3655 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3656 correctly. (thanks to Dan Barlow)
3657 ** portability fixes aiming at bootstrapping under CLISP (thanks
3658 to Dave McDonald and Christophe Rhodes)
3659 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3660 ** fixes in type translation and and type inference (thanks to
3662 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3663 ** various fixes in the new ports (thanks to Dan Barlow)
3664 * several changes related to debugging:
3665 ** suppression of tail recursion, as noted above
3666 ** stack overflow detection, as noted above
3667 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3668 is now the default. (For some time encapsulation has been more
3669 reliable than the breakpoint-based :ENCAPSULATE NIL
3670 implementation, at least on X86 systems; and I just noticed that
3671 encapsulation also seems closer to the spirit of the ANSI
3674 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3676 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3677 up properly again. (There was a packaging bug in 0.7.0 which
3678 left their definitions in SB-SYS::LOAD-FOREIGN and
3679 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3680 vital for most things which interface to C-level interfaces,
3681 like extensions working with sockets or databases or
3682 Perl-compatible regexes or whatever, and the need to fix
3683 this bug is the main reason that 0.7.1 was released so
3685 ** DEFGENERIC is now choosier about the methods it redefines, so that
3686 reLOADing a previously-LOADed file containing DEFGENERICs does
3687 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3688 cycle now works as it should. (thanks to Alexey Dejneka)
3689 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3690 fixed by Christophe Rhodes. (He actually submitted this patch
3691 months ago, and I delayed until after 0.7.0.)
3692 ** Bug 111 (internal compiler confusion about runtime checks on
3693 FUNCTION types) was fixed by Alexey Dejneka.
3694 * Some internal cleanups (getting rid of variables which aren't
3695 needed now that the byte interpreter is gone) caused the fasl
3696 file format number to change again.
3698 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3699 * major incompatible change: The default fasl file extension, i.e. the
3700 default extension for files produced by COMPILE-FILE, has changed
3701 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3703 ** There are many changes in the implementation of the compiler.
3704 SBCL is now essentially a compiler-only implementation of ANSI
3705 Common Lisp. EVAL still "interprets" a few special cases, but
3706 almost all the interesting cases are handled by creating
3707 a LAMBDA expression, calling COMPILE on it, then calling
3708 FUNCALL on the result.
3709 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3710 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3711 Since the code is newer, there might still be some new bugs
3712 (though not as many as before Martin Atzmueller's fixes:-). But
3713 the new code is substantially simpler and clearer, and hopefully
3714 any remaining bugs will be simpler, less fundamental, and more
3715 fixable then the bugs in the old code.
3716 ** The revised compiler is still a little unsteady on its feet.
3718 *** The debugging information it produces (particularly the names
3719 of FUNCTION objects) is sometimes much less useful than what
3720 the old compiler produced.
3721 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3722 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3723 within a MACROLET) has been temporarily weakened.
3724 ** There are new compiler optimizations for various functions:
3725 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3726 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3727 *** the math functions TRUNCATE, FLOOR, and CEILING
3728 *** the function-of-all-trades COERCE
3729 Mostly these should be transparent, but there's one
3730 potentially-annoying problem (bug 117): when the compiler
3731 inline-expands a function and does type analysis on the result,
3732 it can create control paths which have type mismatches, and
3733 when it can't prove that those control paths aren't taken,
3734 it will issue WARNINGs about the type mismatches. This is
3735 a particular problem in practice for the new sequence functions.
3736 It's not clear how this should be fixed, and for now, a
3737 workaround is given in the entry for 117 in the BUGS file.
3738 ** (Because of the interaction between the two previous items --
3739 occasional inlining problems and new inline expansions -- some
3740 of the new sequence function optimizations won't really kick in
3741 completely until debugging information, and then inlining, are
3742 straightened out in some future version.)
3743 * minor incompatible changes:
3744 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3745 DIRECTORY no longer implicitly promotes NIL slots of its
3746 pathname argument to :WILD. In particular, when you ask for the
3747 contents of a directory (which you used to be able to do without
3748 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3749 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3750 ** changes in behavior that ANSI explicitly defines to be
3751 implementation dependent:
3752 *** The new compiler-only implementation still conforms with ANSI,
3753 but acts a little different than before. Besides the obvious
3754 changes in performance tradeoffs (that the cost per form passed
3755 to EVAL has gone up, and the cost per form executed by EVAL
3756 has gone down), the behavior of the system changes a little
3757 because there are no longer any interpreted function objects.
3758 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3759 e.g. doing COMPILE on the output of interactive DEFUN is
3761 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3763 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3764 has changed from (:CL) to NIL.
3765 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3766 e.g. #\Newline and #\Tab, don't have names specified in the
3767 ANSI Common Lisp standard, is now based on their ASCII symbolic
3768 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3769 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3770 are no longer used for output.
3771 ** changes in internal implementation constants:
3772 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3773 4 million. (If your application spends a lot of time GCing and
3774 you have a lot of RAM, you might want to experiment with
3775 increasing it even more.)
3776 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3777 However, almost all old code should still continue to work without
3778 immediate update, as SB-C-CALL is now a (deprecated) nickname
3780 ** Old operator names in the style DEF-FOO are now deprecated in
3781 favor of new corresponding names DEFINE-FOO, for consistency with
3782 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3783 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3784 internal symbols, but a few supported extensions like
3785 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3786 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3787 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3788 etc. as you get deeper into recursive calls to the debugger
3789 command loop, instead of the old "5]", "5]]", "5]]]"
3790 sequence. (I was motivated to do this when squabbles between
3791 ILISP and SBCL left me very deeply nested in the debugger. In the
3792 short term, this change will probably provoke more ILISP/SBCL
3793 squabbles, but hopefully it will be an improvement in the long run.)
3794 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3795 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3796 function, since declaiming return types involving aliens is
3797 (1) annoyingly messy to do by hand and (2) vital to efficient
3798 compilation of code which calls such functions.
3799 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3800 longer reexported by the SB-EXT package. They're solely useful
3801 for alien code, so it seems more logical that you should get
3802 them from the SB-ALIEN package, not in SB-EXT.
3803 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3804 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3805 features. Instead, the code that they used to control is always
3806 built into the system.
3807 * many other bug fixes
3808 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3809 advantage of the new EVAL-WHEN stuff and to clean them up in
3810 general, and they are now more ANSI-compliant in a number of
3811 ways. Martin Atzmueller is responsible for a lot of this.
3812 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3814 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3815 *** correct ERROR type for various file operations
3816 *** some fixes for Lisp streams
3817 *** DEFMETHOD syntax checking
3818 *** changing old weird representation of debug information as
3819 strings (which, among their other deficiencies, don't transform
3820 correctly when you rename packages, and don't change their
3821 print representation when you change things like *PACKAGE*
3822 and *PRINT-LENGTH*) to symbols and lists of symbols
3823 He also made several improvements and fixed several bugs in DESCRIBE.
3824 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3826 *** misbehavior of WRITE-STRING/WRITE-LINE
3827 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3828 and several other LOOP problems as well
3829 *** DIRECTORY when similar filenames are present
3830 *** DEFGENERIC with :METHOD options
3831 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3832 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3833 *** argument ordering in FIND with :TEST option
3834 *** mishandled package designator argument in APROPOS-LIST
3835 *** various problems in the backquote readmacro
3836 *** a bug in APROPOS
3837 *** probably some others that I'm not describing very well here,
3838 since the CVS log documents them by reference to sbcl-devel
3839 messages, and the SourceForge archives aren't working well.:-(
3840 ** Dan Barlow improved the Alpha port (and is making progress on the
3841 PPC port, for those of you who think different).
3842 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3843 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3844 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3845 continues to work on the SPARC port (for those of you in a position
3846 to look down upon our little PC-compatible boxes from a great height).
3847 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3848 and uses that copy, rather than the current dynamic value, when
3849 it's trying to decide whether to truncate output. Thus e.g.
3850 (let ((*print-lines* 50))
3851 (pprint-logical-block (stream nil)
3853 (let ((*print-lines* 8))
3854 (print (aref possiblybigthings i) stream)))))
3855 should now truncate the logical block only at 50 lines, instead of
3856 often truncating it at 8 lines, as it did before.
3857 * The doc/cmucl/ directory, containing old CMU CL documentation
3858 from the time of the fork, is no longer part of the base system.
3859 SourceForge has shut down its anonymous FTP service, and with it
3860 my original plan for distributing the old CMU CL documentation
3861 there. For now, if you need these files you can download an old
3862 SBCL source release and extract them from it.
3863 * The fasl file version number changed again, for dozens of reasons,
3864 some of which are apparent above.
3866 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3867 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3868 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3869 patch, so that DISASSEMBLE output is much nicer.
3870 * The code in the SB-PROFILE package now seems reasonably stable.
3871 I still haven't decided what the final interface should look like
3872 (I'd like PROFILE to interact cleanly with TRACE, since both
3873 facilities use function encapsulation) but if you have a need
3874 for profiling now, you can probably use it successfully with
3875 the current CMU-CL-style interface.
3876 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3877 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3878 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3879 link now returns the dangling link itself, and for similar
3880 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3881 link itself. (In these cases the old code signalled an error and
3882 looped endlessly, respectively.) Thus, DIRECTORY now works even
3883 in the presence of dangling and cyclic symbolic links.
3884 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3885 is now a supported extension again, since the consensus on
3886 sbcl-devel was that it can be useful for ordinary development
3887 work, not just for debugging SBCL itself.
3888 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3889 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3890 that currently-defined functions might be redefined later with
3891 different return types.
3892 * Hash tables can be printed readably, as inspired by CMU CL code
3893 of Eric Marsden and SBCL code of Martin Atzmueller.
3894 * better error handling in CLOS method combination, thanks to
3895 Martin Atzmueller porting Pierre Mai's CMU CL patches
3896 * more overflow fixes for >16Mbyte I/O buffers
3897 * A bug in READ has been fixed, so that now a single Ctrl-D
3898 character suffices to cause end-of-file on character streams.
3899 In particular, now you only need one Ctrl-D at the command
3900 line (not two) to exit SBCL.
3901 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3903 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3904 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3905 does the right thing.
3906 * The implementation of some type tests, especially for CONDITION
3907 types, is now tidier and maybe faster, due to CMU CL code
3908 originally by Douglas Crosher, ported by Martin Atzmueller.
3909 * Some math functions have been fixed, and there are new
3910 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3911 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3912 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3913 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3914 likely to be built into sbcl-0.7.0.)
3915 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3916 CL, ported to SBCL by Martin Atzmueller)
3917 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3918 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3919 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3920 * There's a new slam.sh hack to shorten the edit/compile/debug
3921 cycle for low-level changes to SBCL itself, and a new
3922 :SB-AFTER-XC-CORE target feature to control the generation of
3923 the after-xc.core file needed by slam.sh.
3924 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3925 COMPILE-FILE is no longer supported, so that now every function
3926 gets an entry point, so that block compilation looks a little
3927 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3928 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3929 returns the number of bytes consed since the system started,
3930 rather than the number consed since the first time the function
3931 was called. (The new definition parallels ANSI functions like
3932 CL:GET-INTERNAL-RUN-TIME.)
3933 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3934 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3935 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3936 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3937 * Fasl file version numbers are now independent of the target CPU,
3938 since historically most system changes which required version
3939 number changes have affected all CPUs equally. Similarly,
3940 the byte fasl file version is now equal to the ordinary
3943 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3944 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3945 is no longer recognized. I apologize for this, because it was
3946 listed in SB-EXT as a supported extension, but I found that
3947 its existing behavior was poorly specified, as well as incorrectly
3948 specified, and it looked like too much of a mess to straighten it
3949 out. I have enough on my hands trying to get ANSI stuff to work..
3950 * many patches ported from CMU CL by Martin Atzmueller, with
3951 half a dozen bug fixes in pretty-printing and the debugger, and
3952 half a dozen others elsewhere
3953 * fixed bug 13: Floating point infinities are now supported again.
3954 They might still be a little bit flaky, but thanks to bug reports
3955 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3956 as flaky as they were.
3957 * The --noprogrammer command line option is now supported. (Its
3958 behavior is slightly different in detail from what the old man
3959 page claimed it would do, but it's still appropriate under the
3960 same circumstances that the man page talks about.)
3961 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3962 are now supported, and enabled by default. Thus, the compiler can
3963 handle many floating point and complex operations much less
3964 inefficiently. (Thus e.g. you can implement a complex FFT
3966 * The compiler now detects type mismatches between DECLAIM FTYPE
3967 and DEFUN better, and implements CHECK-TYPE more correctly, and
3968 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3969 thanks to patches from Martin Atzmueller.
3970 * various fixes to make the cross-compiler more portable to
3971 ANSI-conforming-but-different cross-compilation hosts (notably
3972 Lispworks for Windows, following bug reports from Arthur Lemmens)
3973 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3974 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3975 fixed thanks to Pierre Mai's CMU CL patches.
3976 * fixes to keep the system from overflowing internal counters when
3977 it tries to use i/o buffers larger than 16M bytes
3978 * fixed bug 45a: Various internal functions required to support
3979 complex special functions have been merged from CMU CL sources.
3980 (When I was first setting up SBCL, I misunderstood a compile-time
3981 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3982 * improved support for type intersection and union, fixing bug 12
3983 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3984 more obscure bugs as well
3985 * some steps toward byte-compiling non-performance-critical
3986 parts of the system, courtesy of patches from Martin Atzmueller
3987 * Christophe Rhodes has made some debian packages of sbcl at
3988 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3989 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3990 stable, but are nonetheless usable. When he's ready, I'd be happy
3991 to add them to the SourceForge "File Releases" section. (And if
3992 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3993 * new fasl file format version number (because of changes in
3994 internal representation of (OR ..) types to accommodate the new
3995 support for (AND ..) types, among other things)
3997 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3998 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4000 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4002 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4003 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4004 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4005 * DESCRIBE now works on structure objects again.
4006 * Most function call argument type mismatches are now handled as
4007 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4008 know whether the function will be redefined before the call is
4009 executed. (The compiler could flag local calls with full WARNINGs,
4010 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4011 it doesn't keep track of enough information to know whether calls
4012 are local in this sense.)
4013 * Compiler output is now more verbose, with messages truncated
4014 later than before. (There should be some supported way for users
4015 to override the default verbosity, but I haven't decided how to
4016 provide it yet, so this behavior is still controlled by the internal
4017 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4018 src/compiler/ir1util.lisp.)
4019 * Fasl file format version numbers have increased again, because
4020 support for the Gray streams extension changes the layout of the
4021 system's STREAM objects.
4022 * The Gray subclassable streams extension now works, thanks to a
4023 patch from Martin Atzmueller.
4024 * The full LOAD-FOREIGN extension (not just the primitive
4025 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4026 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4027 but like most other programs, it defaults to copying the Unix
4028 environment from the original process instead of starting the
4029 new process in an empty environment.
4030 * Extensions which manipulate the Unix environment now support
4031 an :ENVIRONMENT keyword option which doesn't smash case or
4032 do other bad things. The CMU-CL-style :ENV option is retained
4033 for porting convenience.
4034 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4035 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4037 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4039 * A patch from Martin Atzmueller seems to have solved the SIGINT
4040 problem, and as far as we know, signal-handling now works cleanly.
4041 (If you find any new bugs, please report them!)
4042 * The system no longer defaults Lisp source file names to types
4043 ".l", ".cl", or ".lsp", but only to ".lisp".
4044 * The compiler no longer uses special default file extensions for
4045 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4046 seems to expect a single default extension for all compiled code,
4047 and there's no compelling reason to try to stretch the standard
4048 to allow two different extensions.) Instead, byte-compiled files
4049 default to the same extension as native-compiled files.
4050 * Fasl file format version numbers have increased again, because
4051 a rearrangement of internal implementation packages made some
4052 dumped symbols in old fasl files unreadable in new cores.
4053 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4054 many fewer weird special cases.
4055 * Bug #17 (differing COMPILE-FILE behavior between logical and
4056 physical pathnames) has been fixed, and some related misbehavior too,
4057 thanks to a patch from Martin Atzmueller.
4058 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4059 by Tim Moore, ported to SBCL by Martin Atzmueller.
4060 * Martin Atzmueller fixed several filesystem-related problems,
4061 including bug #36, in part by porting CMU CL patches, which were
4062 written in part by Paul Werkowski.
4063 * More compiler warnings in src/runtime/ are gone, thanks to
4064 more patches from Martin Atzmueller.
4065 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4068 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4070 * DESCRIBE now works on CONDITION objects.
4071 * The debugger now handles errors which arise when trying to print
4072 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4074 * The build system now uses an additional file, customize-target-features.lisp,
4075 to allow local modifications to the target *FEATURES* list. (The point of
4076 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4077 debugging features enabled, without having to worry about propagating it
4078 into everyone's system when I do a "cvs update".) When no
4079 customize-target-features.lisp file exists, the target *FEATURES* list
4080 should be constructed the same way as before.
4081 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4082 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4083 warn about behavior which is undefined under the ANSI spec. Note: This
4084 is specified by ANSI, but it's not very popular with programmers.
4085 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4086 macro in the SBCL sources for an example of a workaround which you
4087 might use to make portable ANSI-standard code which does what you want.
4088 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4089 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4090 when it's used as a non-toplevel form. (This is required in order
4091 to implement the DEFCONSTANT-EQX macro.)
4092 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4093 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4094 funny things, due to bugs in the implementation of EVAL-WHEN
4095 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4096 non-toplevel DEFCONSTANTs are uncommon.)
4097 * The core file version number and fasl file version number have been
4098 incremented, because the old noncompliant DEFCONSTANT behavior involved
4099 calling functions which no longer exist, and because I also took the
4100 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4101 * fixed bug 1 (error handling before read-eval-print loop starts), and
4102 redid debugger restarts and related debugger commands somewhat while
4104 ** The QUIT debugger command is gone, since it did something
4105 rather different than the SB-EXT:QUIT command, and since it never
4106 worked properly outside the main toplevel read/eval/print loop.
4107 Invoking the new TOPLEVEL restart provides the same functionality.
4108 ** The GO debugger command is also gone, since you can just invoke
4109 the CONTINUE restart directly instead.
4110 ** The TOP debugger command is also gone, since it's redundant with the
4111 FRAME 0 command, and since it interfered with abbreviations for the
4113 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4115 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4116 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4117 some porting work by Martin Atzmueller)
4118 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4119 with a boatload of patches which he ported to SBCL. Now that those
4121 ** The system tries to make sure that its low-priority messages
4122 are prefixed by semicolons, to help people who like to use
4123 syntax highlighting in their ILISP buffer. (This patch
4124 was originally due to Raymond Toy.)
4125 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4126 patches originally written by Raymond Toy.
4127 ** The compiler understands coercion between single-value and
4128 multiple-VALUES type expressions better, getting rid of some very
4129 weird behavior, thanks to patches originally by Robert MacLachlan
4130 and Douglas Crosher.
4131 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4132 lambda lists, thanks to a patch originally by Pierre Mai.
4133 ** The system no longer bogusly warns about "abbreviated type
4135 ** The compiler gets less confused by inlining and RETURN-FROM,
4136 thanks to some patches originally by Tim Moore.
4137 ** The system no longer hangs when dumping circular lists to fasl
4138 files, thanks to a patch originally from Douglas Crosher.
4139 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4140 undefined function error.
4141 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4142 in structure slot initforms) for now, documented workaround instead:-|
4143 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4144 * fixed bug 5: added stubs for various Gray stream functions called
4145 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4146 installed, at least appropriate type errors are generated
4147 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4148 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4149 consistently in DEFMETHOD forms.
4150 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4151 it doesn't seem to affect SBCL after all
4152 * The C runtime system now builds with better optimization and many
4153 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4155 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4157 * The system is now under CVS at SourceForge (instead of the
4158 CVS repository on my home machine).
4159 * The new signal handling code has been tweaked to treat register
4160 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4161 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4162 downstream, e.g. in debug-int.lisp, has implicit dependencies
4163 on the unsignedness of integer representation of machine words,
4164 and that caused the system to bomb out with infinite regress
4165 when trying to recover from type errors involving signed values,
4166 e.g. (BUTLAST '(1 2 3) -1).)
4167 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4168 (This was one of the bugs Peter Van Eynde reported back in July.)
4169 * The system now uses code inspired by Colin Walters' O(N)
4170 implementation of MAP (from the cmucl-imp@cons.org mailing
4171 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4172 inline the MAP operation, and there is more than one
4173 sequence argument to the MAP call (so that it can't just
4174 do ETYPECASE once and for all based on the type of the
4175 single sequence argument). (The old non-inline implementation
4176 of the general M-argument sequence-of-length-N case required
4177 O(M*N*N) time when any of the sequence arguments were LISTs.)
4178 * The QUIT :UNIX-CODE keyword argument has been renamed to
4179 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4180 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4182 * Raymond Wiker's patches to port dynamic loading from Linux to
4183 FreeBSD have been added.
4184 * The BUGS file is now more nearly up to date, thanks in large part
4185 to Martin Atzmueller's review of it.
4186 * The debugger now flushes standard output streams before it begins
4187 its output ("debugger invoked" and so forth).
4188 * The core version number and fasl file version number have both
4189 been incremented, because of incompatible changes in the layout
4191 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4192 used to not be called for a saved Lisp image.)
4193 * Martin Atzmueller's version of a patch to fix a compiler crash,
4194 as posted on sbcl-devel 13 September 2000, has been installed.
4195 * Instead of installing Martin Atzmueller's patch for the
4196 compiler transform for SUBSEQ, I deleted the compiler transform,
4197 and transforms for some similar consing operations.
4198 * A bug in signal handling which kept TRACE from working on OpenBSD
4200 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4201 away by constant folding
4202 * The system now defines its address space constants in one place
4203 (in the Lisp sources), and propagates them automatically elsewhere
4204 (through GENESIS and the sbcl.h file). Therefore, patching the
4205 address map is less unnecessarily tedious and error-prone. The
4206 Lisp names of address space constants have also been systematized.
4207 * CVS tags like dollar-Header-dollar have been removed from
4208 the sources, because they have never saved me trouble and
4209 they've been source of trouble working with patches and other
4210 diff-related operations.
4211 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4212 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4214 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4216 * The system has been ported to OpenBSD.
4217 * The system now compiles with a simple "sh make.sh" on the systems
4218 that it's supported on. I.e., now you no longer need to tweak
4219 text in the target-features.lisp-expr and symlinks in src/runtime/
4220 by hand, the make.sh takes care of it for you.
4221 * The system is no longer so grossly inefficient when compiling code
4222 involving vectors implemented as general (not simple) vectors (VECTOR T),
4223 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4224 which dares to use the various sequence functions on non-simple
4225 vectors, takes less of a performance hit.
4226 * There is now a primitive type predicate VECTOR-T-P
4227 to test for the (VECTOR T) type, so that e.g.
4228 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4229 can now be compiled with some semblance of efficiency. (The old code
4230 turned the type declaration into a full call to %TYPEP at runtime!)
4231 * AREF on (VECTOR T) is still not fast, since it's still compiled
4232 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4233 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4234 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4235 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4236 without making full calls to SUBTYPEP at runtime.
4237 (Some analogous efficiency issues for non-simple vectors specialized to
4238 element types other than T, or for non-simple multidimensional arrays,
4239 have not been addressed. They could almost certainly be handled the
4240 same way if anyone is motivated to do so.)
4241 * The changes in array handling break binary compatibility, so
4242 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4243 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4244 instead of (VALUES T T).
4245 * By following the instructions that Dan Barlow posted to sbcl-devel
4246 on 2 July 2000, I was able to enable primitive dynamic object
4247 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4248 functionality is not implemented (since it calls ld to resolve
4249 library references automatically, requiring RUN-PROGRAM for its
4250 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4251 try to resolve library references) is now supported.
4252 * The system now flushes the standard output streams when it terminates,
4253 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4254 them at several other probably-convenient times, e.g. in each pass of
4255 the toplevel read-eval-print loop, and after evaluating a form given
4256 as an "--eval" command-line option. (These changes were motivated by a
4257 discussion of stream flushing issues on cmucl-imp in August 2000.)
4258 * The source transform for TYPEP of array types no longer assumes
4259 that an array whose element type is a not-yet-defined type
4260 is implemented as an array of T, but instead punts, so that the
4261 type will be interpreted at runtime.
4262 * There is now some support for cross-compiling in make.sh: each of
4263 the phases of make.sh has its own script. (This should be transparent
4264 to people doing ordinary, non-cross-compile builds.)
4265 * Since my laptop doesn't have hundreds of megabytes of memory like
4266 my desktop machine, I became more motivated to do some items on
4267 my to-do list in order to reduce the size of the system a little:
4268 ** Arrange for various needed-only-at-cold-init things to be
4269 uninterned after cold init. To support this, those things have
4270 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4271 symbols with such names are now uninterned after cold init).
4272 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4273 fasl files for cold load.
4274 ** Remove the old compiler structure pooling code (which used to
4275 be conditional on the target feature :SB-ALLOC) completely.
4276 ** Redo the representation of some data in cold init to be more compact.
4277 (I also looked into supporting byte compiled code at bootstrap time,
4278 which would probably reduce the size of the system a lot, but that
4279 looked too complicated, so I punted for now.)
4280 * The maximum signal nesting depth in the src/runtime/ support code has
4281 been reduced from 4096 to 256. (I don't know any reason for the very
4282 large old value. If the new smaller value turns out to break something,
4283 I'll probably just bump it back up.)
4284 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4286 * Many, many bugs reported by Peter Van Eynde have been added to
4287 the BUGS list; some have even been fixed.
4288 * While enabling dynamic object file loading, I tried to make the
4289 code easier to understand, renaming various functions and variables
4290 with less ambiguous names, and changing some function calling
4291 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4292 * While trying to figure out how to do the OpenBSD port, I tried to
4293 clean up some of the code in src/runtime/. In particular, I dropped
4294 support for non-POSIX signal handling, added various comments,
4295 tweaked the code to reduce the number of compilation warnings, and
4296 renamed some files to increase consistency.
4297 * To support the new automatic configuration functionality in make.sh,
4298 the source file target-features.lisp-expr has been replaced with the
4299 source file base-target-features.lisp-expr and the machine-generated
4300 file local-target-features.lisp-expr.
4301 * fixed a stupid quoting error in make.sh so that using CMU CL
4302 "lisp -batch" as cross-compilation host works again
4304 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4306 * DESCRIBE no longer tries to call itself recursively to describe
4307 bound/fbound values, so that it no longer fails on symbols which are
4308 bound to themselves (like keywords, T, and NIL).
4309 * DESCRIBE now works on generic functions.
4310 * The printer now prints less-screwed-up representations of closures
4311 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4312 * A private symbol is used instead of the :EMPTY keyword previously
4313 used to mark empty slots in hash tables. Thus
4314 (DEFVAR *HT* (MAKE-HASH-TABLE))
4315 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4316 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4317 now does what ANSI says that it should. (You can still get
4318 similar noncompliant behavior if bang on the hash table
4319 implementation with all the symbols you get back from
4320 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4321 This breaks binary compatibility, since tests for equality to
4322 :EMPTY are wired into things like the macroexpansion of
4323 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4325 * There's now a minimal placeholder implementation for CL:STEP,
4326 as required by ANSI.
4327 * An obscure bug in the interaction of the normal compiler, the byte
4328 compiler, inlining, and structure predicates has been patched
4329 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4330 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4332 * Missing ordinary arguments in a macro call are now detected even
4333 when the macro lambda list contains &KEY or &REST.
4334 * The debugger no longer complains about encountering the top of the
4335 stack when you type "FRAME 0" to explicitly instruct it to go to
4336 the top of the stack. And it now prints the frame you request even
4337 if it's the current frame (instead of saying "You are here.").
4338 * As specified by ANSI, the system now always prints keywords
4339 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4341 * The default initial SIZE of HASH-TABLEs is now smaller.
4342 * Type information from CLOS class dispatch is now propagated
4343 into DEFMETHOD bodies, so that e.g.
4344 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4346 is now basically equivalent to
4347 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4348 (DECLARE (TYPE SINGLE-FLOAT X))
4350 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4351 operation, without having to do run-time type dispatch.
4352 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4353 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4354 * Since I don't seem to be making big file reorganizations very often
4355 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4356 is overflowing my ability to conveniently back them up), I've finally
4357 checked the system into CVS. (The CVS repository is on my home system,
4358 not at SourceForge -- putting it on SourceForge might come later.)
4359 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4360 high-level GC-NOTIFY-FOO functions send their output. (There's
4361 still very little control of where low-level verbose GC functions
4362 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4363 less than it used to -- the GC-NOTIFY-FOO functions are now under
4364 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4365 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4366 in only one place in the source code, and propagates it automatically
4367 everywhere that it's needed. Thus e.g. when I bump the version from
4368 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4369 * The C source files now include boilerplate legalese and documentation
4370 at the head of each file (just as the Lisp source files already did).
4371 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4372 to his page will be replaced with a link to his new CLiki service.
4374 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4376 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4377 * The build process now looks for GNU make under the default name "gmake",
4378 instead of "make" as it used to. If GNU make is not available as "gmake"
4379 on your system, you can change this default behavior by setting the
4380 GNUMAKE environment variable.
4381 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4382 can build without error under CMU CL.
4384 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4386 * There is now a partial SBCL user manual (with some new text and some
4387 text cribbed from the CMU CL manual).
4388 * The beginnings of a profiler have been added (starting with the
4389 CMU CL profiler and simplifying and cleaning up). Eventually the
4390 main interface should be through the TRACE macro, but for now,
4391 it's still accessed through vaguely CMU-CL-style functions and macros
4392 exported from the package SB-PROFILE.
4393 * Some problems left over from porting CMU CL to the new
4394 cross-compilation bootstrap process have been cleaned up:
4395 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4396 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4397 part when getting rid of the compiler *BACKEND* stuff.)
4398 ** The value of *NULL-TYPE* was screwed up, because it was
4399 being initialized before the type system knew the final
4400 definition of the 'NULL type. This screwed up several key
4401 optimizations in the compiler, causing inefficiency in all sorts
4402 of places. (I found it because I wanted to understand why
4403 GET-INTERNAL-RUN-TIME was consing.)
4404 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4405 PROCLAIM FTYPE information. Unfortunately this broke binary
4406 compatibility again, since now the forms output by DEFGENERIC
4407 to refer to functions which didn't exist in 0.6.3.
4408 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4409 can use the new (as of 0.6.3) transform for SOME into MAP into
4411 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4412 values to (INTEGER 0 1000000), so that the time code will no longer
4413 occasionally get blown up by Linux returning 1000000 microseconds
4414 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4415 its output conform to the ANSI spec. (Alas, this makes its output
4416 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4417 * A full call to MAP NIL with a single sequence argument no longer conses.
4418 * fixes to problems pointed out by Martin Atzmueller:
4419 * The manual page no longer talks about multiprocessing as though
4420 it were currently supported.
4421 * The ILISP support patches have been removed from the distribution,
4422 because as of version 5.10.1, ILISP now supports SBCL without us
4423 having to maintain patches.
4424 * added a modified version of Raymond Toy's recent CMU CL patch for
4425 EQUALP comparison of HASH-TABLE
4427 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4429 * The system still can't cross-compile itself with
4430 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4431 entails), but at least it can compile more of itself that way
4432 than it used to be able to, and various buglets which were uncovered
4433 by trying to cross-compile itself that way have now been fixed.
4434 * This release breaks binary compatibility again. This time
4435 at least I've incremented the FASL file format version to 2, so that the
4436 problem can be detected reliably instead of just causing weird errors.
4437 * various new style warnings:
4438 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4439 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4440 ** using the lexical binding of a variable named in the *FOO* style
4441 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4442 as specified by ANSI.
4443 * *RANDOM-STATE* is no longer automatically initialized from
4444 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4445 default behavior of the system is to repeat its behavior every time
4446 it's run. If you'd like to change this behavior, you can always
4447 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4448 old convention there was no comparably easy way to get the system to
4449 repeat its behavior every time it was run.
4450 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4451 FTYPE declarations has been removed, in favor of their ANSI
4452 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4453 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4454 the types of their sequence arguments can be determined at compile time.
4455 This is done through a new open code expansion for MAP which eliminates
4456 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4457 argument types can be determined at compile time.
4458 * The optimizer now transforms COERCE into an identity operation when it
4459 can prove that the coerced object is already of the correct type. (This
4460 can be a win for machine generated code, including the output of other
4461 optimization transforms, such as the MAP transform above.)
4462 * Credit information has been moved from source file headers into CREDITS.
4463 * Source file headers have been made more standard.
4464 * The CASE macro now compiles without complaining even when it has
4467 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4469 * (Note that the way that the PCL macroexpansions were rewritten
4470 to accommodate the change in DEFGENERIC below breaks binary
4471 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4472 not run under sbcl-0.6.2. Once we get out of alpha releases,
4473 i.e. hit release 1.0.0, we'll probably try to maintain binary
4474 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4475 and sbcl-1.4.4. Until then, however, it might be fairly common
4476 for maintenance releases to break binary compatibility.)
4477 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4479 * The reporting of unhandled signals has been changed to print some
4480 explanatory text as well as the report form. (Previously only
4481 the report form was printed.)
4482 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4483 it defines, so that the compiler no longer issues undefined function
4484 warnings for compiled-but-not-yet-loaded generic functions.
4485 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4486 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4487 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4488 section of the ANSI spec are supported.
4489 * The "" nickname for the "KEYWORD" package has been removed.
4490 The reader still handles symbol tokens which begin with a package marker
4491 as keywords, but it doesn't expose its mechanism for doing so in the
4492 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4493 * The system now issues STYLE-WARNINGs for contradictory TYPE
4494 proclamations. (Warnings for contradictory FTYPE proclamations would
4495 be nice too, but those can't be done usefully unless the type system
4496 is made smarter about FUNCTION types.)
4497 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4498 systematized, so that "*target-*.lisp is supposed to exist only on the
4499 target and imply that there's a related file which exists on the
4500 host, and *host-*.lisp is supposed to exist only on the host and imply
4501 that there's a related file which exists on the target. This involves a
4502 lot of renaming. Hopefully the acute confusion caused by the renaming
4503 will be justified by the reduction in chronic confusion..
4504 ** runtime-type.lisp -> early-target-type.lisp
4505 ** target-type.lisp -> late-target-type.lisp
4506 ** early-host-format.lisp -> early-format.lisp
4507 ** late-host-format.lisp -> late-format.lisp
4508 ** host-error.lisp -> misc-error.lisp
4509 ** early-error.lisp -> early-target-error.lisp
4510 ** late-error.lisp -> late-target-error.lisp
4511 ** host-defboot.lisp -> early-defboot.lisp
4512 ** code/misc.lisp -> code/target-misc.lisp
4513 ** code/host-misc.lisp -> code/misc.lisp
4514 ** code/numbers.lisp -> code/target-numbers.lisp
4515 ** code/early-numbers.lisp -> numbers.lisp
4516 ** early-host-type.lisp -> early-type.lisp
4517 ** late-host-type.lisp -> late-type.lisp
4518 ** host-typep.lisp -> typep.lisp
4519 ** load.lisp -> target-load.lisp
4520 ** host-load.lisp -> load.lisp
4521 ** host-disassem.lisp -> disassem.lisp
4522 ** host-insts.lisp -> insts.lisp
4523 ** byte-comp.lisp -> target-byte-comp.lisp
4524 ** host-byte-comp.lisp -> byte-comp.lisp
4525 ** host-signal.lisp -> signal.lisp
4526 ** host-defstruct.lisp -> defstruct.lisp
4527 ** late-target-type.lisp -> deftypes-for-target.lisp
4528 Furthermore, several other previously target-only files foo.lisp (e.g.
4529 hash-table.lisp and random.lisp) have been split into a target-and-host
4530 foo.lisp file and a target-only target-foo.lisp file, with their key type
4531 definitions in the target-and-host part, so that the cross-compiler will
4532 know more about target types.
4533 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4534 gone away. In their place are various *BACKEND-FOO* variables
4535 corresponding to the slots of the old structure.
4536 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4537 into the target SBCL has been fixed.
4538 * The chill.lisp system for loading cold code into a running SBCL
4540 * Support for the CMU CL "scavenger hook" extension has been removed.
4541 (It was undocumented and unused in the CMU CL sources that SBCL was
4542 derived from, and stale in sbcl-0.6.1.)
4543 * Various errors in the cross-compiler type system were detected
4544 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4545 (enabling various consistency checks). Many of them were fixed,
4546 but some hard problems remain, so the compiler is back to
4547 running without *TYPE-SYSTEM-INITIALIZED* for now.
4548 * As part of the cross-compiler type system cleanup, I implemented
4549 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4550 * I have started adding UNCROSS calls throughout the type system
4551 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4552 on cross-compiler input files will be able to go away, and various
4554 * CONSTANTP now returns true for quoted forms (as explicitly required
4557 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4559 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4560 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4561 (MAKE-LIST -1). In the longer term, it ought to become true
4562 that declarations are assertions even at SAFETY 1. For now, it's not
4563 quite true even at SAFETY 3, but it's at least more nearly true..
4564 (Note that this change seems to increases the size of the system by
4565 O(5%) and to decrease the speed of the compiler by 20% or more.)
4566 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4567 for the problem of printing dozens of lines of distracting information
4568 about low-level system machinery as part of the top stack frame
4569 on entry to the debugger when an undefined function was called.
4570 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4571 is not reset to COMMON-LISP-USER.
4572 * Compilation of stuff related to dyncount.lisp has been made conditional
4573 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4574 smaller. The various dyncount-related symbols have been moved into
4575 a new "SB-DYNCOUNT" package.
4576 * tty-inspect.lisp has been renamed to inspect.lisp.
4577 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4578 feature has gone away. (When we eventually port to other flavors of
4579 libc and/or Unix, we'll try to make the differences between flavors
4580 invisible at the user level.)
4581 * Various other *FEATURES* tags, and/or their associated conditionals,
4582 have been removed if obsolescent, or given better documentation, or
4583 sometimes given more-mnemonic names.
4585 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4587 * tidied up "make.sh" script
4588 * tidied up system directory structure
4589 * better "clean.sh" behavior
4590 * added doc/FOR-CMUCL-DEVELOPERS
4591 * many many small tweaks to output format, e.g. removing possibly-confusing
4592 trailing #\. character in DESCRIBE-INSTANCE
4593 * (EQUALP #\A 'A) no longer signals an error.
4594 * new hashing code, including EQUALP hashing
4595 * tidied up Lisp initialization and toplevel
4596 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4597 * command line argument processing
4598 * added POSIX-GETENV function to deal with Unix-ish environment variables
4599 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4600 terminating SBCL on EOF
4601 * non-verbose GC by default
4602 * There is no more "sbcl" shell script; the sbcl file is now the C
4603 runtime executable (just like CMU CL).
4604 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4606 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4607 symbol and package information as Lisp native symbol and package objects
4608 instead of strings naming symbols and strings naming packages. This way,
4609 whenever packages are renamed (as in warm init), debug information is
4610 transformed along with everything else.
4611 * tweaked the optimization policy declarations which control the building
4612 of SBCL itself. Now, among other things, the system no longer saves
4613 source location debugging information. (This helps two problems at once
4614 by reducing SBCL size and by keeping SBCL from trying to look for its
4615 sources -- which may not exist -- when reporting errors.)
4616 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4617 debugging and testing purposes
4618 * cleaned up printing, making the printer call PRINT-OBJECT for
4619 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4620 methods, giving nearly-ANSI behavior
4621 * converted almost all special variables to use *FOO* naming convention
4622 * deleted PARSE-TIME functionality, since it can be done portably
4623 * moved some files out of cold init into warm init
4624 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4626 * regularized formatting of source files
4627 * added an install.sh script
4628 * fixed ridiculous memory usage of cross-compiler by making
4629 compiler/alloc.lisp not try to do pooling unless it can hook
4630 itself into the GC of the cross-compilation host. Now the system
4631 builds nicely on my old laptop.
4632 * added :SB-ALLOC in target-features.lisp-expr
4633 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4634 was not implemented)
4635 * re-did condition handling and note reporting in the compiler. Notes
4636 are no longer handled by signalling conditions. Style warnings
4637 and warnings are handled more correctly and reported in such a way
4638 that it's easy to find one or the other in your output (so that you
4639 can e.g. figure out which of many problems caused COMPILE-FILE to
4641 * changed the severity of several compiler warnings from full WARNING
4642 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4643 compiler note reporting so that it doesn't use the condition system
4644 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4645 COMPILE-FILE command)
4646 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4647 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4648 no longer supported, since their implementation was deeply intertwingled
4649 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4650 * removed bogus "support" for compiler macros named (SETF FOO), and
4651 removed the compiler macro for SETF INFO (but only after making a fool
4652 of myself on the cmucl-imp mailing list by posting a bogus patch for
4653 DEFINE-COMPILER-MACRO..)
4654 * Compiled files containing forms which have side effects on the Lisp
4655 reader (such as DEFPACKAGE forms) are now handled more correctly.
4656 (Compiler queuing of top level lambdas has been suppressed by setting
4657 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4658 * deleted various currently-unused source files, e.g. gengc.lisp. They
4659 may be added back at some point e.g. when porting to other architectures,
4660 but until they are it's distracting to distribute them and to try to
4662 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4663 there were so many of them they're just distractions, and UNCROSS is
4664 known to be able to handle the current sources
4665 * moved PROFILE functionality into TRACE, so that it will be clear
4666 how the wrapping and unwrapping of functions when you profile them
4667 interacts with the wrapping and unwrapping of functions when you
4668 trace them. (Actually, the functionality isn't there yet, but at least
4669 the interface specification is there. Hopefully, the functionality will
4670 arrive with some future maintenance release.)
4671 * removed host-oops.lisp
4672 * changed signature of QUIT function to allow UNIX-CODE argument
4673 * fixed READ-SEQUENCE bug
4674 * tweaked verbose GC output so that it looks more like the progress
4675 output that ANSI specifies for functions like LOAD
4676 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4677 * added <http://sbcl.sourceforge.com> to the banner information printed by