Alexey Dejneka [Wed, 11 Jun 2003 05:53:59 +0000 (05:53 +0000)]
0.8.0.61:
* bug fix: in macro-like defining macros/special operators the
implicit block does not enclose lambda list.
Christophe Rhodes [Tue, 10 Jun 2003 13:04:32 +0000 (13:04 +0000)]
0.8.0.60:
Fix yet another Dietz bug:
... in DEFCLASS, we only redefine an existing class with name
CLASS-NAME if said name is the proper name for the class;
... ask for a new class (by passing NIL to ENSURE-CLASS-USING-CLASS)
otherwise.
Christophe Rhodes [Tue, 10 Jun 2003 11:08:09 +0000 (11:08 +0000)]
0.8.0.59:
Make VALUES derive type optimizer accurate:
... since we have this nice interpretation of VALUES types, use
it in the VALUES derive-type optimizer -- we will return
exactly as many values as VALUES has arguments;
... fix the bugs that this reveals in sbcl; :-)
... enables us to detect more bogosity: test for some more type
mismatches being caught.
... (relatedly) fix one more duplicate definition in
sb-bsd-sockets
Array initializer type warning fix:
... don't do (csubtypep (ctype-of x) eltype), because that's wrong
for e.g. X = #\a and eltype being STANDARD-CHAR; use
CTYPEP instead.
Christophe Rhodes [Tue, 10 Jun 2003 08:28:23 +0000 (08:28 +0000)]
0.8.0.58:
Minor fixes:
... don't define a LENGTH type for disassembly; we're not
allowed to do that, because LENGTH is in the CL package. Call
it DISASSEM-LENGTH instead. (reported by rtoy on #lisp IRC)
... cut'n'pasteo: restart print functions in warm init now refer
to variables that are bound, rather than those that aren't.
Alexey Dejneka [Tue, 10 Jun 2003 06:48:57 +0000 (06:48 +0000)]
0.8.0.57:
* Signal a style warning when DECLAIM is met in a declaration
position;
* Don't join blocks if the separating continuation's dest is CRETURN;
* DO-USES: in the restart mode stop iterations when the block
is deleted under us.
William Harold Newman [Mon, 9 Jun 2003 20:33:56 +0000 (20:33 +0000)]
0.8.0.56:
another way to break the compiler -- seems to be a bad day
for interaction of my coding style with SBCL
Christophe Rhodes [Mon, 9 Jun 2003 17:05:53 +0000 (17:05 +0000)]
0.8.0.55:
Contrib fixes for non-x86/non-linux
... don't use GET-LISP-OBJ-ADDRESS on fixnums or characters;
... don't try to define "posix" linux-specific functions
also minor frobs
Christophe Rhodes [Mon, 9 Jun 2003 16:21:16 +0000 (16:21 +0000)]
0.8.0.54:
COMPILE-FILE and "static linking"
... we are allowed to consider references to functions defined
in the same file as such. At present, we don't inline such
references, but merely use previously-derived type information
when compiling calls.
... also, since the consequences are undefined for multiple
definitions in the same file, add a warning for that case (and
fix the examples in the codebase itself :-)
William Harold Newman [Mon, 9 Jun 2003 15:19:24 +0000 (15:19 +0000)]
0.8.0.53:
new BUGS entry
Daniel Barlow [Sun, 8 Jun 2003 21:30:28 +0000 (21:30 +0000)]
0.8.0.52
large contrib/simple-streams patch including many CMUCL
changes by Paul Foley
(Thanks to Rudi Schlatte sbcl-devel 2003.06.08)
... melded streams are implemented
... removed device-extend; it's deprecated in acl's
simple-streams implementation and will go away there too
... prettier print-object methods
... various small fixes
... Implemented file-position
... Removed some sbcl internals redefinition: no need to
rewrite unix-lseek, use sb-posix for mmap, munmap
... Can now remove sb-grovel requirement too
... Commented out (but not deleted yet) heavily
sbcl-internals-dependent slot access machinery
(def-stream-class, with-stream-class, sm).
... Various bug fixes, e.g. read-line now works for lines
longer than 80 characters
Daniel Barlow [Sun, 8 Jun 2003 21:24:15 +0000 (21:24 +0000)]
0.8.0.52
large contrib/simple-streams patch including many CMUCL
changes by Paul Foley
(Thanks to Rudi Schlatte sbcl-devel 2003.06.08)
... melded streams are implemented
... removed device-extend; it's deprecated in acl's
simple-streams implementation and will go away there too
... prettier print-object methods
... various small fixes
... Implemented file-position
... Removed some sbcl internals redefinition: no need to
rewrite unix-lseek, use sb-posix for mmap, munmap
... Can now remove sb-grovel requirement too
... Commented out (but not deleted yet) heavily
sbcl-internals-dependent slot access machinery
(def-stream-class, with-stream-class, sm).
... Various bug fixes, e.g. read-line now works for lines
longer than 80 characters
Christophe Rhodes [Sun, 8 Jun 2003 18:43:53 +0000 (18:43 +0000)]
0.8.0.51:
Fix TYPE-OF bugs from Paul Dietz' test suite
... attempt to return reasonable intersections of the relevant
built-in-types for non-negative integers
Christophe Rhodes [Sun, 8 Jun 2003 15:02:27 +0000 (15:02 +0000)]
0.8.0.50:
Fixes for ADD-METHOD.[12] and FIND-METHOD error cases
... ADD-METHOD should return the generic function (but preserve
method-returning in the internal function ADD-NAMED-METHOD
... FIND-METHOD needs to signal an error if the lengths of the
specializers isn't the same as the number of required arguments
to the generic function. Turn the test in REAL-GET-METHOD into
an AVER.
... REMOVED-NAMED-METHOD is unused; delete it.
... incompatible lambda lists don't actually require an error of
type PROGRAM-ERROR to be signalled, and in fact this change
can make the error signalled be an ERROR. Adjust the test.
Alexey Dejneka [Sun, 8 Jun 2003 09:30:01 +0000 (09:30 +0000)]
0.8.0.49:
Added support of disassembling of closures and funcallable
instances.
Alexey Dejneka [Sun, 8 Jun 2003 06:12:30 +0000 (06:12 +0000)]
0.8.0.48:
Fixed type declarations for functions returning trees.
Daniel Barlow [Sat, 7 Jun 2003 22:46:08 +0000 (22:46 +0000)]
0.8.0.47
sb-simple-streams pushes :little-endian/:big-endian on
*features* for no apparent reason - and gets it wrong, too
unix-lseek is fixed, so sb-simple-streams need not redefine
it any more
Fix stupid braino in unix-ioctl that broke the build
Christophe Rhodes [Sat, 7 Jun 2003 16:37:23 +0000 (16:37 +0000)]
0.8.0.46:
Fix for argumentless CALL-NEXT-METHOD and assignment
... and a simple test.
Daniel Barlow [Sat, 7 Jun 2003 15:52:03 +0000 (15:52 +0000)]
0.8.0.45
64 bit alien support on x86, by reference to CMUCL
... x86/sap.lisp gets transforms for *-sap-ref-64 which
do the reference a word at a time
... teach define-alien-type-method (integer :extract-gen)
to know about *-sap-ref-64
... #!+alpha removed from defknowns for *-sap-ref-64
... compiler/x86/c-call gets hairy great deftransform %alien-funcall
that transforms 64 bit args and return values
appropriately. Lifted & ported from CMUCL by rude mechanicals
This all could use generifying for endianness and putting
somewhere that all 32 bit platforms (i.e. everything but
Alpha, I think) can see it.
Fix the BSD-specific mess in unix-lseek to use off-t like
the other platforms do. Should now work (untested)
Add mmap/munmap/getpagesize support to SB-POSIX,
courtesy Rudi Schlatte.
SB-UNIX:UNIX-IOCTL second argument is signed, as per
sbcl-devel mail from Vincent Arkesteijn
Christophe Rhodes [Sat, 7 Jun 2003 15:09:22 +0000 (15:09 +0000)]
0.8.0.44:
Fix amazingly horrendous bug in constant-dfun generation (thanks
to Gerd Moellmann)
... when the method combination isn't standard, method functions
might not be effective methods.
... in constant-value-miss, aver that we are finding a
constant-value.
... test case from Paul Dietz
Christophe Rhodes [Sat, 7 Jun 2003 13:49:11 +0000 (13:49 +0000)]
0.8.0.43:
Fix bug 63, by propagating symbol macros into the walker's
variable environment
... no, I don't claim to have understood the walker fully;
... test for correct behaviour in sb-cltl2's MACROEXPAND-ALL.
Christophe Rhodes [Sat, 7 Jun 2003 11:01:58 +0000 (11:01 +0000)]
0.8.0.42:
Quieten the compiler slightly for (LENGTH <VECTOR>)
... one of the transforms wants to constant-fold the above when
<VECTOR> is a 1d SIMPLE-ARRAY of known dimensions.
However, the deftransform definition (with a simple-array
type requirement) caused the compiler to complain when
the argument was known to be a vector but not a simple-array.
Instead, leave the type requirement for the transform open,
and GIVE-UP-IR1-TRANSFORM (with no arguments => silence)
if the argument isn't simple.
... write some tests for LENGTH on vectors
Alexey Dejneka [Sat, 7 Jun 2003 06:08:58 +0000 (06:08 +0000)]
0.8.0.41: "Hunting on worms, part II"
Insertion of type check on a function result continuation
prevents tail call optimisation, so:
* pathwise eliminate unnecessary type checks;
* disable unsafe CAST insertion on function results.
Now Paul Dietz' test suite finishes with "83 out of 12565
total tests failed".
Christophe Rhodes [Fri, 6 Jun 2003 12:52:52 +0000 (12:52 +0000)]
0.8.0.40:
Fix build on non-x86 (confirmed for sparc; probable for the
others...)
... we can't WRITE (or FORMAT) as early in cold-init as we
used to, because *PRINT-PPRINT-DISPATCH* is now unbound for
typecheck reasons. Create the FLOAT-REG-SYMBOL vectors at
read-time instead.
Alexey Dejneka [Fri, 6 Jun 2003 09:09:12 +0000 (09:09 +0000)]
0.8.0.39:
* Convert proclaimed function result type into a type check.
... recanned several incautious worms.
Christophe Rhodes [Thu, 5 Jun 2003 21:07:10 +0000 (21:07 +0000)]
0.8.0.38:
Fix build with :SB-SHOW (patch from Nikodemus Siivola sbcl-devel
2003-05-31)
... also some whitespace (and now-outdated comment) deletion
in target-error
Christophe Rhodes [Thu, 5 Jun 2003 19:58:32 +0000 (19:58 +0000)]
0.8.0.37:
Make DEFCONSTANT throw an error of type SB-EXT:DEFCONSTANT-UNEQL
... rearrange build order so that macros are defined at the
birth of the cross-compiler (and that relevant variables are
declared special)
... define the condition type (slots for OLD-VALUE, NEW-VALUE
and NAME)
... actually throw the error in %DEFCONSTANT (and have an ABORT
restart, too)
... document and test
Alexey Dejneka [Thu, 5 Jun 2003 17:25:27 +0000 (17:25 +0000)]
0.8.0.36:
Included module SB-CLTL2.
Christophe Rhodes [Thu, 5 Jun 2003 15:52:56 +0000 (15:52 +0000)]
0.8.0.35:
Compiler depessimization, (part IV of XL :-)
... let's see what happens if we're a little less demure about
putting derived function types into the globaldb. We don't
actually use them unless SB-EXT:*DERIVE-FUNCTION-TYPES* is true,
so this shouldn't make a difference to the compiler in normal
circumstances (though it does when the compiler is being built,
and also can do for libraries who choose to use the extension)
This is probably a staging post on the way to ANSI-style
optimization of file-compilation, wherein we can treat all the
functions in one file as tightly coupled (see CLHS 3.2.2.3)
Alexey Dejneka [Thu, 5 Jun 2003 06:24:26 +0000 (06:24 +0000)]
0.8.0.34:
* Fixed bug 253.
* "Exactly" declare types of alien routines.
Daniel Barlow [Wed, 4 Jun 2003 15:29:38 +0000 (15:29 +0000)]
0.8.0.33
Extend RUN-PROGRAM with a :SEARCH option that makes it look
along $PATH for the executable
Export SB-EXT:FIND-EXECUTABLE-IN-SEARCH-PATH, which does the
actual lookup in the preceding function
Add a couple of test cases
Alexey Dejneka [Wed, 4 Jun 2003 13:39:32 +0000 (13:39 +0000)]
0.8.0.32:
Put CSR's bug report about type checking in CAST chains.
Alexey Dejneka [Wed, 4 Jun 2003 06:03:09 +0000 (06:03 +0000)]
0.8.0.31:
Optional entry points are converted on demand.
Alexey Dejneka [Wed, 4 Jun 2003 05:58:18 +0000 (05:58 +0000)]
IR1-convertion of lambda is separated into
src/compiler/ir1tran-lambda.lisp.
Daniel Barlow [Tue, 3 Jun 2003 22:34:53 +0000 (22:34 +0000)]
0.8.0.30
Update ASDF from upstream:
... export OPERATION-ON-{WARNINGS,FAILURE}
... add verbosity control to make loading optionally much
more quiet (operate 'load-op :verbose nil)
... muffle STYLE-WARNINGS from loaded code in REQUIRE hook.
Perhaps slightly contentious, but I think while it's fair
to present these to developers, they just clutter the place
up as far as end-users are concerned
Some TODO notes in SB-POSIX
Christophe Rhodes [Tue, 3 Jun 2003 16:03:43 +0000 (16:03 +0000)]
0.8.0.29:
Port fix to PCL from Gerd Moellmann regarding metacircles:
there was a hole in the metacircularity detection of legacy PCL,
which is filled by bypassing ordinary slot access for standard
classes in cache miss handling when doing so would lead to
another cache miss.
... *CACHE-MISS-VALUES-STACK* and *STANDARD-SLOT-LOCATIONS* (not
audited for threadsafety)
Christophe Rhodes [Tue, 3 Jun 2003 14:57:59 +0000 (14:57 +0000)]
0.8.0.28:
Efficient version of WITH-RECURSIVE-LOCK
... since the frame-pointer is lispobj aligned, we can safely
pun it into a fixnum without fearing that we'll point
into space. All those SAP-INTs and MAKE-LISP-OBJs then
compile into null moves, and no more efficiency notes.
Alexey Dejneka [Tue, 3 Jun 2003 12:29:16 +0000 (12:29 +0000)]
0.8.0.27:
Fixed bug 252: use UNLINK-NODE to delete returns in merging
toplevel lambdas.
William Harold Newman [Mon, 2 Jun 2003 22:12:04 +0000 (22:12 +0000)]
0.8.0.27:
redid DESCRIBE and DESCRIBE-OBJECT newlining/freshlining to
be consistent with each other (so e.g. we don't get
multiple leading newlines when DESCRIBEing an instance
of STRUCTURE-OBJECT) and to conform to ANSI spec with
minimal surprise:
...DEFUN DESCRIBE now longer FRESH-LINEs.
...DEFUN DESCRIBE no longer does PPRINT-LOGICAL-BLOCK, either,
since FRESH-LINE inside PP-L-B can make a mess.
...DESCRIBE-OBJECT methods consistently do FRESH-LINEs, as in
the spec example, and if they use the prettyprinter,
they create their own PPRINT-LOGICAL-BLOCKs.
(No, this style -- coders paid by the line, mixing high level
CLOS dispatch with low-level physical output bypassing
the pretty-printer -- is not the way that I would have
specified the behavior, but I was still programming in C
and C++ when the spec was written, and no one asked me.)
deleted *DESCRIBE-METAOBJECTS-AS-OBJECTS-P*, since its output
is so messy I doubt people want to use it (and if I'm
wrong the implementation is trivial to restore, with
the only trickiness being figuring out a decent
interface to support)
added warning for the unwary/unwise in SB-BSD-SOCKETS docs
Christophe Rhodes [Mon, 2 Jun 2003 15:55:48 +0000 (15:55 +0000)]
0.8.0.26:
A couple of contrib fixes
... adjust the dependencies in sb-posix (patch from Rudi
Schlatte sbcl-devel 2003-06-02)
... gray-streams stream-foo-sequence take their arguments in
the other order (David Lichteblau 2003-06-01)
Christophe Rhodes [Mon, 2 Jun 2003 15:41:33 +0000 (15:41 +0000)]
0.8.0.25:
Merge "(defmacro nada (()))" patch (David Lichteblau sbcl-devel
2003-06-02)
Christophe Rhodes [Sun, 1 Jun 2003 15:44:54 +0000 (15:44 +0000)]
0.8.0.24:
A couple of small fixes:
... make the raw-slot reffer/setters in the case of
UNSIGNED-BYTE raw type (when the accessor is just AREF) know
that they're dealing with a (SIMPLE-ARRAY (UNSIGNED-BYTE 32)
(*)), quietening the compiler a little.
... make a fallback constructor, not an optimized constructor,
for condition classes. (reported by eightjean on #lisp IRC)
Christophe Rhodes [Sun, 1 Jun 2003 11:15:54 +0000 (11:15 +0000)]
0.8.0.23:
Fix a couple of bugs from Paul Dietz' test suite
... multiple class/superclass redefinitions can leave a layout
with an INVALID value of T; this case needs to be handled in
CHECK-WRAPPER-VALIDITY. (thanks to Gerd Moellmann)
... (SETF FIND-CLASS) will eventually pass NIL to (SETF
FIND-CLASSOID), so that had better be able to handle it.
Christophe Rhodes [Sun, 1 Jun 2003 10:48:21 +0000 (10:48 +0000)]
0.8.0.22:
Unbreak WITH-PACKAGE-ITERATOR (which I managed to break when
introducing enough type declarations to quieten the compiler)
... in the :INHERITED case, ,',SYMBOLS isn't necessarily true,
so when ,',HASH-VECTOR isn't bound to a hash-vector, we
need to set ,',COUNTER to NIL, rather than try seeking
down ,',HASH-VECTOR.
Daniel Barlow [Sat, 31 May 2003 20:56:02 +0000 (20:56 +0000)]
0.8.0.21
Add contrib/asdf-module.mk to the list of files that go into a
binary distribution (thanks to David Lichteblau for bug report)
Christophe Rhodes [Fri, 30 May 2003 11:26:58 +0000 (11:26 +0000)]
0.8.0.20:
Minor contrib infrastructure frob:
... don't use :force t to ensure building, because firstly it's
hideously expensive in time, as we have to build base systems
several times; secondly, it won't work if the semantics change
to not propagate the FORCE value to depended systems; and
thirdly it doesn't work anyway in the presence of our current
TEST-OP implementation. Delete suspicious files (including
*.fasl under contrib/) at the start of make-target-contrib.sh
instead.
... cvsignore contrib/systems
Christophe Rhodes [Fri, 30 May 2003 10:44:10 +0000 (10:44 +0000)]
0.8.0.19:
No-one's complained, so merge CSR patch "Type checking on global
variables" (sbcl-devel 2003-05-27)
... fix SB-XC:PROCLAIM to queue up TYPE and FTYPE proclamations
when the system isn't initialized, and then reproclaim them
later;
... fix EVAL to punt to the compiler if there's a type proclamation
for FOO in (SETQ FOO ...);
... proclaim the types of the various CL:*FOO* variables, according
to the CLHS;
... fix two instances of undefined behaviour in the test suite. :-)
Christophe Rhodes [Fri, 30 May 2003 09:39:21 +0000 (09:39 +0000)]
0.8.0.18:
Fix bug in APROPOS (reported by cliini on #lisp IRC 2003-05-29)
... it's the second value from FIND-SYMBOL that we want to
compare to :EXTERNAL, not the primary :)
... while we're at it, quieten WITH-PACKAGE-ITERATOR (and
incidentally LOOP FOR ... BEING EACH SYMBOL IN ...)
Daniel Barlow [Fri, 30 May 2003 02:35:48 +0000 (02:35 +0000)]
0.8.0.17
SB-GROVEL : include stdio.h in generated C file
unconditionally - we use printf, after all
SB-POSIX specification updates: more detail on types,
designators, foreign memory
SP-POSIX TODO list updated with feedback from c.l.l thread
Christophe Rhodes [Thu, 29 May 2003 16:14:44 +0000 (16:14 +0000)]
0.8.0.16:
Code deletion, yay
... since we've expunged *BYTES-CONSED-BETWEEN-GCS* everywhere else,
we might as well stop it being set and confusing people, so
delete it;
... since our CMUCL brethren have decided that 12Mb is a good value
for (BYTES-CONSED-BETWEEN-GCS), let's follow that. Note that
this increases the default on x86, but decreases the historical
default on non-x86 (though in recent times a refactor had
likewise changed that default);
... fix an error caught by new function warning stuff: bogus
parenthesis in a sparc VOP.
Christophe Rhodes [Thu, 29 May 2003 12:28:01 +0000 (12:28 +0000)]
0.8.0.15:
A couple more minor fixes:
... LOOP FOR ... FROM ... can apparently accept complex numbers
in some cases. Ew. Make it so, but attempt to limit
the damage by still providing compile-time diagnostics
where possible.
... disassemble FUCOM on x86 correctly. (thanks to Raymond Toy)
... unBAshify test script. (thanks to Henrik Motakef)
William Harold Newman [Wed, 28 May 2003 15:15:23 +0000 (15:15 +0000)]
0.8.0.14:
changed #(1.2.3.4) value (changed in 0.8.0.12) in the sources
too (with more guidance from ZB on #lisp)
Christophe Rhodes [Wed, 28 May 2003 14:49:35 +0000 (14:49 +0000)]
0.8.0.13:
Miscellaneous grab bag of fixes
... use SB!XC:MOST-POSITIVE-FIXNUM in bit-bash type; the
expression is no doubt still wrong for 64-bit lisps, but
it stands a chance of being right for 32-bit; :)
... NIL is a valid structure slot name; (yes, really!)
... whine about invalid keywords in macro calls even if there
are no defined keywords (but just &KEY) in the lambda
list;
... prettify the compile-time warning in
%COMPILE-TIME-TYPE-ERROR a little;
... a couple of IGNOREs
William Harold Newman [Wed, 28 May 2003 14:26:48 +0000 (14:26 +0000)]
0.8.0.12:
added .cvsignore files in contrib/, and made corresponding
changes in clean.sh (didn't clean up foo.c in clean.sh)
(Maybe if we called it grovel-tmp.c or something I'd
be more comfortable with its treewide autodeletion.)
fixed #(1.2.3.4) thinko in sb-bsd-sockets docs (pointed out
by Zachary Beane on sbcl-devel 2003-05-27)
Christophe Rhodes [Wed, 28 May 2003 11:51:10 +0000 (11:51 +0000)]
0.8.0.11:
Merge patch from Antonio Martinez (sbcl-devel 2003-05-19
"read-sequence for fundamental-binary-input-streams")
... and WRITE-SEQUENCE, as well.
Merge patch from Andreas Fuchs (sbcl-devel 2003-05-20
"(defmethod foo (&key bla &rest blub) t)")
... detect more erroneous specialized lambda lists;
... handle the error in the SOURCE-CONTEXT method for DEFMETHOD,
otherwise we can't print the error message we want;
... fix the bogus lambda list in simple-streams and in the
test suite;
... include tests for a variety of bogus input.
(I would also like to mark the occasion of SBCL's very own millennium
bug, as we observe the CVS revision number for version.lisp-expr tick
over from 1.999 to 1.1000. "Should auld acquaintance be forgot...")
Alexey Dejneka [Tue, 27 May 2003 18:13:39 +0000 (18:13 +0000)]
0.8.0.10:
Beautified (I hope) error reporting for "single-value"
compile-time type errors.
Christophe Rhodes [Tue, 27 May 2003 16:17:26 +0000 (16:17 +0000)]
0.8.0.9:
Minor buglet fixes:
... remove bogus YES-OR-NO-P and Y-OR-N-P extra level of
indirection. (thanks to Antonio Martinez)
... make clocc-ansi-test whine less, by not making
COMPILER-ERROR inherit from SERIOUS-CONDITION
... add a couple of IGNORE/IGNORABLEs
Christophe Rhodes [Tue, 27 May 2003 13:32:57 +0000 (13:32 +0000)]
0.8.0.8:
Some slight MAKE-LOAD-FORM-related fixes
... in general, slots can be named by any symbols; DEFCLASS is
more stringent in its requirements, so move the extra
checks into the DEFCLASS macro.
... now structure slots can be named by keywords again.
... make MAKE-LOAD-FORM-SAVING-SLOTS results on structures
cause the compiler to be less verbose, by using a
lower-level setter (SB!KERNEL:SLOT-SETTER-LAMBDA-FORM).
[ oops, forgot to commit the tests ]
Christophe Rhodes [Tue, 27 May 2003 13:32:11 +0000 (13:32 +0000)]
0.8.0.8:
Some slight MAKE-LOAD-FORM-related fixes
... in general, slots can be named by any symbols; DEFCLASS is
more stringent in its requirements, so move the extra
checks into the DEFCLASS macro.
... now structure slots can be named by keywords again.
... make MAKE-LOAD-FORM-SAVING-SLOTS results on structures
cause the compiler to be less verbose, by using a
lower-level setter (SB!KERNEL:SLOT-SETTER-LAMBDA-FORM).
Christophe Rhodes [Tue, 27 May 2003 09:20:01 +0000 (09:20 +0000)]
0.8.0.7:
Various contrib/ fixes
... isolate the sb-bsd-sockets tests from the rest of the
system, so that they're not included in the
sb-simple-streams tests. (thanks to Rudi Schlatte)
... replace asm/errno.h with errno.h include in
sb-simple-streams. (thanks to Rudi Schlatte)
... make sb-grovel more likely to emit C9x compliant code, by
minimizing rightward drift.
Alexey Dejneka [Tue, 27 May 2003 08:35:52 +0000 (08:35 +0000)]
0.8.0.6:
* Walker knows about NAMED-LAMBDA;
* implemented short form of VALUES type specifier.
Alexey Dejneka [Mon, 26 May 2003 14:42:23 +0000 (14:42 +0000)]
0.8.0.5:
* Deleted obsolete bug 244;
* optimizer for ARRAY-HEADER-P knows about 0-dimensional
arrays;
... fixed bug 250.
Alexey Dejneka [Mon, 26 May 2003 08:17:13 +0000 (08:17 +0000)]
0.8.0.4:
Fixed bug 249: local functions did not check type of unused
arguments.
Alexey Dejneka [Mon, 26 May 2003 04:25:52 +0000 (04:25 +0000)]
0.8.0.3:
Merged CAST branch.
Changes since -cast.8:
* separated usage of object and values types;
* fixed warning and error reports for compile-time type
errors;
* inline structure slot accessors are implemented with source
transforms;
* enabled warning emitting for type errors in some paths to
CAST;
* removed check for type errors in arguments of a call of a
flushable function;
* source transforms are made nameless.
Christophe Rhodes [Sun, 25 May 2003 22:34:23 +0000 (22:34 +0000)]
0.8.0.2:
Fix stack exhaustion stack exhaustion death
... define DEFINE-FUNCTION-NAME-SYNTAX function-name-defining macro;
... use it for SETF functions, and define LEGAL-FUNCTION-NAME-P
and FUN-NAME-BLOCK-NAME in terms of VALID-FUNCTION-NAME-P;
... also define internal PCL generalized function name syntax as
such, and test for internalness in SET-ARG-INFO1;
... OAOO bonus: delete bits of SB!PCL::CLASS-PREDICATE that were
decorating the compiler;
(note: this API is interface-compatible with CMUCL's for defining
generalized function name syntax. However, it's not currently exported
from SB-EXT because I happen to think that calling something
VALID-FUNCTION-NAME-P when it returns two values, the second of which
is syntactically significant, is a bit lame, and maybe we'll be able
to agree a better name between the two projects)
Christophe Rhodes [Sun, 25 May 2003 22:26:13 +0000 (22:26 +0000)]
0.8.0.1:
Fix build on FreeBSD
... don't #include <proc.h>, which defines a struct thread; (thanks
to Henrik Motakef)
... do sigaltstack() if #+c-stack-is-control-stack (which is, in
practice, all the time).
(this /may/ break building on OpenBSD, which uses the same OS-dependent
routines; it is possible that OpenBSD needs something from <proc.h>;
if so, we need to find some compromise)
William Harold Newman [Sun, 25 May 2003 15:11:35 +0000 (15:11 +0000)]
0.8.0:
release, tagged as sbcl_0_8_0
Christophe Rhodes [Thu, 22 May 2003 16:46:58 +0000 (16:46 +0000)]
0.8alpha.0.45:
Fix profiler bug reported by APD sbcl-devel 2003-05-21:
... update *N-BYTES-FREED-OR-PURIFIED* in SUB-GC;
... add a smoke test.
Increment FASL file version number due to change in
DEFSTRUCT-SLOT-DESCRIPTION structure.
Kevin Rosenberg [Thu, 22 May 2003 16:00:20 +0000 (16:00 +0000)]
0.8alpha.0.44:
sb-aclrepl/inspect.lisp: Change usage of dsd-%name to dsd-name
Kevin Rosenberg [Tue, 20 May 2003 16:15:18 +0000 (16:15 +0000)]
0.8alpha.0.43:
contrib/sb-aclrepl:
Rework newline handling to ensure a fresh line
Correct README file
Christophe Rhodes [Tue, 20 May 2003 14:42:25 +0000 (14:42 +0000)]
0.8alpha.0.42:
improvements to sb-simple-streams contrib (from Rudi Schlatte)
... don't spam *features* anymore
... writes of large chunks of data work now, instead of
failing silently
Christophe Rhodes [Tue, 20 May 2003 10:49:26 +0000 (10:49 +0000)]
0.8alpha.0.41:
Firefighting the build, part II
... remove DSD-%NAME optimization, in the interest of making
SLOT-VALUE (and hence MAKE-LOAD-FORM-SAVING-SLOTS)
a more reliable operation.
... now the name of the slot is the symbol in the DEFSTRUCT
form, as expected; also, now the CL package is pristine,
containing only the 978 exported symbols.
Essentially this version has built from CMUCL and built itself
successfully.
Christophe Rhodes [Tue, 20 May 2003 10:36:02 +0000 (10:36 +0000)]
0.8alpha.0.40:
Firefighting the build, part I
... make SLOT-VALUE work on :READ-ONLY T structure slots
Notes:
* This version may have a subtle breakage that may or may not
bite. The forthcoming 0.8alpha.0.41 commit will fix that
subtle breakage, but it was fixed chronologically before in my
tree, so I haven't taken out that fix to test this one in
isolation. I hope that makes sense...)
* The new implementation of MAKE-LOAD-FORM-SAVING-SLOTS seems
very slow; this may be a perceptual problem. What is
incontrovertible is that it is very noisy; it chatters about
compiling many top-level forms, caused by PCL generating
LOAD-TIME-VALUE forms for ENSURE-ACCESSOR to optimize SLOT-VALUE.
A fix for this, at least for structure objects, will probably be
forthcoming in 0.8alpha.0.4x.
Christophe Rhodes [Mon, 19 May 2003 16:25:10 +0000 (16:25 +0000)]
0.8alpha.0.39:
A couple more CLOS fixes:
... make &OPTIONAL argument count checking less lax in methods
(caught by pfdietz' MAKE-LOAD-FORM.ERROR.2)
... make :ARGUMENT-PRECEDENCE-ORDER and :METHOD-COMBINATION
DEFGENERIC options do sanity checking on their arguments
(:A-P-O caught by pfdietz' suite; :M-C
checking defensively installed :-)
Christophe Rhodes [Mon, 19 May 2003 14:47:14 +0000 (14:47 +0000)]
0.8alpha.0.38:
Slight change to REQUIRE/PROVIDE protocol
... as observed by Tony Martinez sbcl-devel 2003-05-13, REQUIRE
takes a string designator, so allow this
... update to latest ASDF, which changes the asdf hook slightly
such that individual modules are required (ha!) to
PROVIDE themselves; make it so.
Christophe Rhodes [Mon, 19 May 2003 14:05:17 +0000 (14:05 +0000)]
0.8alpha.0.37:
Make MAKE-LOAD-FORM(-SAVING-SLOTS) vaguely conform
... and in the process, remind myself of just how horrible the
:JUST-DUMP-IT-NORMALLY hack was.
... more methods on MAKE-LOAD-FORM;
... real, CLOS-based introspective definition of
MAKE-LOAD-FORM-SAVING-SLOTS...
... which means that we have to hold off from using MLFSS until
it's around, so make JUST-DUMP-IT-NORMALLY use
:SB-JUST-DUMP-IT-NORMALLY rather than MLFSS in its
definition for the target.
Fix the type.impure.lisp test for the new definition of
condition classes (oops).
Christophe Rhodes [Mon, 19 May 2003 10:51:32 +0000 (10:51 +0000)]
0.8alpha.0.36:
A little tender loving care, applied to conditions:
... make the implementation of DEFINE-CONDITION agree with the
documentation string: allow :DOCUMENTATION slot options
to work.
... ANSI (and pfdietz :-) wants SLOT-EXISTS-P to work on
conditions; hook condition objects into CLOS enough to
talk about existence of slots: (new classes
CONDITION-{EFFECTIVE,DIRECT}-SLOT-DEFINITION,
CONDITION-CLASS, etc)
... it's a bit ridiculous to have SLOT-EXISTS-P working on
conditions, and then not be able to do SLOT-VALUE, so
do the work necessary to make CONDITION objects
more-or-less fully understood by PCL: (new methods on
COMPUTE-SLOTS, ALLOCATE-INSTANCE, SLOT-VALUE-USING-CLASS
and friends; new clauses in internal functions such as
GET-OPTIMIZED-STD-SLOT-VALUE-USING-CLASS-METHOD-FUNCTION;
adjustment of the braid to set up CLOS knowledge of the
new class hierarchy).
Christophe Rhodes [Sun, 18 May 2003 15:49:11 +0000 (15:49 +0000)]
0.8alpha.0.35:
Commit patch from Rudi Schlatte for sb-simple-streams contrib
... add TODO
... implement socket streams (using sb-bsd-sockets)
... (frob by CSR: don't fail if we get a connection refused)
Christophe Rhodes [Sun, 18 May 2003 14:42:33 +0000 (14:42 +0000)]
0.8alpha.0.34:
Make ALLOCATE-INSTANCE work on all structure classes
... if we're defined by a DEFSTRUCT, then make a closure to
allocate an instance.
Christophe Rhodes [Fri, 16 May 2003 14:42:35 +0000 (14:42 +0000)]
0.8alpha.0.33:
Add one more contrib
... SB-MD5, based on Pierre Mai's highly-frobbed implementation of
RFC1321;
... a couple of cosmetic changes to rotate-byte to accommodate the
demands of code that wants to use it.
Christophe Rhodes [Fri, 16 May 2003 11:37:09 +0000 (11:37 +0000)]
0.8alpha.0.32:
Improvements in build technology:
... detect and choose GNUMAKE earlier, once and only once.
... provide recompile and use-anyway restarts for warm load.
Kevin Rosenberg [Thu, 15 May 2003 16:21:41 +0000 (16:21 +0000)]
0.8alpha.0.31:
* sb-aclrepl/sb-aclrepl.asd: Remove compilation note workaround
* sb-aclrepl/repl.lisp: add a few needed newlines to output
Christophe Rhodes [Thu, 15 May 2003 16:03:33 +0000 (16:03 +0000)]
0.8alpha.0.30:
Bandage over compiler overenthusiasm for EQL-specialized methods
... reduction in annoyance for contrib/ users probably more
important than informing developers that they should
be using EQL on SINGLE-FLOAT, not T, for best
performance.
Christophe Rhodes [Wed, 14 May 2003 14:16:41 +0000 (14:16 +0000)]
0.8alpha.0.29:
A couple of extremely minor fixes
... "oops" in type declaration in %DEFUN
... fix KLUDGE in IMMEDIATE-CONSTANT-SC for x86, and allow (in
principle) building from lisps with larger
MOST-POSITIVE-FIXNUM as well as smaller.
Christophe Rhodes [Tue, 13 May 2003 13:55:29 +0000 (13:55 +0000)]
0.8alpha.0.28:
Fix bug 47d (DEFGENERIC must signal PROGRAM-ERROR when
attempting to create a generic function with the same name as a
special operator).
... sounds easy, huh? No.
... make COMPILER-ERROR not inherit from ERROR any more, so that
user handlers don't (wrongly) claim to handle it;
... establish a handler for COMPILER-ERROR around the evaluator
that delegates to the compiler handlers if present, but
handles them itself if not...
... by signalling an error from a new internal restart, to allow
user handlers for ERROR and friends a chance to run.
Christophe Rhodes [Tue, 13 May 2003 12:21:15 +0000 (12:21 +0000)]
0.8alpha.0.27:
Add skeletal simple-streams contrib from Rudi Schlatte, based on Paul
Foley's implementation for CMUCL.
Kevin Rosenberg [Mon, 12 May 2003 18:12:35 +0000 (18:12 +0000)]
0.8alpha.0.26:
* sb-contrib/sb-aclrepl.asd:
- Remove work-around for optimization notes
* sb-contrib/{toplevel,repl}.lisp:
- Rework fresh-line handling to accomodate that *repl-read-fun* causes
a newline of which the output-stream is unaware.
Christophe Rhodes [Mon, 12 May 2003 14:10:30 +0000 (14:10 +0000)]
0.8alpha.0.25:
A couple of PCL fixes:
... REMOVE-METHOD should always return its generic function
argument. Make it so.
... SHARED-INITIALIZE should initialize :CLASS slots too.
Kevin Rosenberg [Mon, 12 May 2003 04:40:30 +0000 (04:40 +0000)]
0.8alpha.0.24:
* sb-aclrepl/sb-aclrepl.asd:
- Work around 'eql method specialization optimization notes
* sb-aclrepl:repl.lisp:
- Refactor read-cmd into small functions
- Add relative history numbers, eg, `:-2'
- Add history pattern match search, eg, `::foo'
- Add optional redo query to history command, eg, `:24 ?'
Christophe Rhodes [Fri, 9 May 2003 10:22:48 +0000 (10:22 +0000)]
0.8alpha.0.23:
Fix a couple of the CLOS bugs that have been accumulating:
... make :ALLOCATION :CLASS slots behave as they should in slot
inheritance and class redefinition.
Slightly unKLUDGEify the %SET-SYMBOL-VALUE implementation
... do it the same way whether building with #!+SB-THREAD or
not, so at least it's ugly once and only once.
Daniel Barlow [Thu, 8 May 2003 20:37:54 +0000 (20:37 +0000)]
0.8alpha.0.22
SB-GROVEL-related changes
... make sb-bsd-sockets use it
... which involved restoring size-of-{int,char,long} constants
When building contrib, reset the central-registry to avoid
picking up any code from e.g. ~/.sbcl/systems/
Christophe Rhodes [Thu, 8 May 2003 16:17:57 +0000 (16:17 +0000)]
0.8alpha.0.21:
The "uh, I thought we had users beta-testing for us" release:
Make SETQ on globals return a value (and IN-PACKAGE, and ...)
... the SET VOP and the %SET-SYMBOL-VALUE IR2 thingy are
different, so one shouldn't translate the other;
... instead, create an IR2 translator for %SET-SYMBOL-VALUE by
hand, mimicking the effect of :SET-TRANS in
DEFINE-PRIMITIVE-OBJECT;
... this removes the need for late-symbol.lisp, so delete it.
While we're in a deleting mood...
... delete src/pcl/fast-init.lisp, which hasn't been part of the
build for the last 5 months or so.
Christophe Rhodes [Thu, 8 May 2003 15:43:21 +0000 (15:43 +0000)]
0.8alpha.0.20:
Make OpenMCL a suitable slamming host:
... fix slam.sh
... :if-exists :supersede for *output-files-for-genesis*
Since OpenMCL builds still work, announce clisp as a suitable host
for the cross-compiler.
Daniel Barlow [Thu, 8 May 2003 12:44:18 +0000 (12:44 +0000)]
0.8alpha.0.19
We're not using ALTERNATE-SIGNAL-STACK-START, and in fact
haven't been for some time. Remove it.
Christophe Rhodes [Thu, 8 May 2003 11:17:24 +0000 (11:17 +0000)]
0.8alpha.0.18:
Some minor package alterations:
... export CONTROL-STACK-POINTER-VALID-P from SB!VM;
... along with new *{BINDING,CONTROL}-STACK-FOO* to replace old
non-asterisked versions;
make CONTROL-STACK-POINTER-VALID-P not style-warn on each use.
Daniel Barlow [Wed, 7 May 2003 11:38:35 +0000 (11:38 +0000)]
0.8alpha.0.17
Package frobbing fix from Matthew Danish to make sb-thread
build again
Save errno return from waitpid long enough to print the error,
as the real errno is clobbered by parent_do_garbage_collect()
Christophe Rhodes [Wed, 7 May 2003 11:18:59 +0000 (11:18 +0000)]
0.8alpha.0.16:
Fix for SXHASH on condition objects
... was causing compilation failures when referencing explicit
constant conditions
William Harold Newman [Wed, 7 May 2003 02:42:04 +0000 (02:42 +0000)]
0.8alpha.0.15:
broke some long lines
merged Matthew Danish's slam-on-clisp patch from sbcl-devel
Daniel Barlow [Mon, 5 May 2003 23:27:07 +0000 (23:27 +0000)]
0.8alpha.0.14
Merge thread-gc-branch.
Summary: move time-to-gc-p logic entirely into C. Delete a
lot of Lisp stuff no longer necessary. Make SUB-GC
thread-safe or at least thread-tolerant. Some hooks and
variables that were previously available but not apparently
used for much are now no longer present.
Christophe Rhodes [Mon, 5 May 2003 14:42:08 +0000 (14:42 +0000)]
0.8alpha.0.13:
(oops: add file needed from last commit)
Christophe Rhodes [Mon, 5 May 2003 14:09:03 +0000 (14:09 +0000)]
0.8alpha.0.13:
CLISP build megapatch
... mostly putting #-SB-XC in front of :COMPILE-TOPLEVEL,
because clisp gives a full warning for function and
macro redefinition;
... workaround clisp's buggy pretty printer by not exercising it
as much: use (INHIBIT-WARNINGS 3);
... explicit :INITIAL-ELEMENT 0 when we're using 0 to mean
"uninitialized" in MAKE-ARRAY;
... SPECIAL-OPERATOR-P isn't a good test on the host for what
can become a target macro;
... slightly more portable floating point logic:
Explicitly set *READ-DEFAULT-FLOAT-FORMAT* so that we
don't create host LONG-FLOATs by accident;
LOAD-TIME-VALUE magic for negative floating point zeros;
Minor associated text file frobbage
... braindump some unrelated TODO items
Obligatory runtime code improvement
... fix one warning in gencgc.h