Commits on Jun 23, 2002
  1. 0.7.5:

    William Harold Newman committed Jun 23, 2002
Commits on Jun 22, 2002

    	Reopened BUG 140, closing BUG 176 in the process.
    csrhodes committed Jun 22, 2002
Commits on Jun 20, 2002

    	finishing up the immediately preceding "BUGS and other text" checkin:-|
    William Harold Newman committed Jun 20, 2002
  2. BUGS and other text

    William Harold Newman committed Jun 20, 2002

    	SPARC floating point fixes
    	... write a C function to get at the floating point state register
    		and use it for context-floating-point-modes (SunOS)
    	... attempt to do the same for SPARC/Linux, then realise that
    		the current state was more broken than I thought, so
    		wrote a BUG instead
    	Portability fix to
    csrhodes committed Jun 20, 2002
Commits on Jun 18, 2002

    	merged pmai patch (sbcl-devel 2002-06-18) for bug 180 (where
    		:MOST-SPECIFIC-LAST options were ignored in method
    William Harold Newman committed Jun 18, 2002

    	Final OpenMCL fixes
    	... dubiously fix a FIXME in the reader with something that
    		superficially looks better but is in fact just as
    	Now SBCL compiles under an unpatched released OpenMCL!
    csrhodes committed Jun 18, 2002

    	Some cleanups for OpenMCL compilation
    	... change one IGNORE to IGNORABLE
    	... delete FIXME from errors (OpenMCL can compile that now)
    	... NB: it still doesn't quite work for some reason;
    		investigation is ongoing...
    	Text file updates
    csrhodes committed Jun 18, 2002
Commits on Jun 16, 2002

    	just BUGS entries
    William Harold Newman committed Jun 16, 2002
Commits on Jun 15, 2002

    	(no real changes, just note of another cvs commit mistake:
    		not making a log notice of the NARROWED-EMF code in
    		boot.lisp code which replaces the fix for
    		bug 177. The newer code, which seems to've gone in
    		in, endeavors to give the compiler enough
    		information to delete the code which was causing type
    		warnings, instead of the .24 hack of just hiding the
    		offending code behind an opaque identity function)
    William Harold Newman committed Jun 15, 2002

    	minor OAOO FIXME for GENESIS...
    	..made the FOO-ENTRY-TYPE-CODE parameters propagate
    		automatically into sbcl.h instead of maintaining
    		separate copies of their definititions in core.h (and
    		renamed them, s/entry-type-code/core-entry-type-code/,
    		 to help make them more painfully specific now that
    		their scope is wider)
    	...similarly propagated FOO-SPACE-ID automatically into
    		sbcl.h, and s/foo-space-id/foo-core-space-id/
    William Harold Newman committed Jun 15, 2002
Commits on Jun 14, 2002

    	The missing piece in the OpenMCL build...
    	... move some clauses around in cross-compilation float logic.
    		NB: I am _not_ sure that this is correct in any sense
    		other than the empirical "it works". The IEEE-representation
    		logic for the cross-compiler needs review by someone
    		who knows what an IEEE float looks like. For now, though...
    	... now SBCL builds under OpenMCL!
    csrhodes committed Jun 14, 2002

    	pure lossage: Only an elite few languages limit their built-in
    		"vector" abstraction to short fixed lengths, but Common
    		Lisp is one of them, so in order to use vectors in
    		GENESIS with OpenMCL as xc host, we get to roll our own
    		implementation of longer vectors. (and fair warning: If
    		in the next few months anyone dares to suggest a
    		correlation between Lisp and "the Right Thing" and I
    		don't detect enough sarcasm to stun a Cape Buffalo at
    		fifty paces, I plan to kick him until I feel better.)
    		But on what some might consider to be the plus side,
    		this is a case where it's easy to keep the
    		comment-to-code ratio around 1:1...
    William Harold Newman committed Jun 14, 2002

    	mostly comment cleanups, but also a few slot renamings from my
    		(unsuccessful so far) attempts to understand why a test
    		case makes PROPAGATE-LIVE-TNS chew up 95% of compiler
    	... s/global-conflicts-next/global-conflicts-next-blockwise/
    	... s/global-conflicts-tn-next/global-conflicts-next-tnwise/
    	... (i.e. making parallel-in-meaning slots parallel in name)
    William Harold Newman committed Jun 14, 2002
Commits on Jun 13, 2002

    	Merge BYTE fix.
    	... include LDB-TEST in the shadowed symbols
    	... leave hideous violation of OAOO in load-or-cload-xcompiler
    		unfixed for now
    csrhodes committed Jun 13, 2002

    	merged patch: APD ANSI compliance bugfix for DECLARE IGNORABLE
    		of special variable (sbcl-devel 2002-06-12)
    	Whether or not my alternative fix for the CALL-NEXT-METHOD
    		type warning bug (being carefully reviewed on
    		sbcl-devel even as I write:-) is OK, I might as well
    		at least stuff the exploratory test cases I wrote
    		for it into tests/.
    	Bug 137 (useless debug names e.g. in BACKTRACE) seems to be
    		gone, since at least I haven't stumbled across it
    		recently, so delete the BUGS entry.
    William Harold Newman committed Jun 13, 2002
Commits on Jun 12, 2002

    	Fix BUG 179
    	... one character fix, and two-line test.
    	Some OpenMCL cross-compilation fixes
    	... prefer (declaim (ftype (*) ...)) to (declare (values ...))
    	... remove arrays that are undumpable from cross-compiler's logic.
    csrhodes committed Jun 12, 2002

    	OpenMCL bootstrap fix
    	... (INTEGER-LENGTH x) is the same as (1+ (FLOOR (LOG x 2))), duh,
    		and is anyway what I was trying to express.
    	Log the bug in DIRECTORY
    csrhodes committed Jun 12, 2002
Commits on Jun 11, 2002

    	new BUGS entry
    William Harold Newman committed Jun 11, 2002

    	Bugfix for PPC floating point control handling
    	... document and test.
    	Minor text file tweaking, too
    csrhodes committed Jun 11, 2002

    	Fix bug 177 (bogus type warnings from CALL-NEXT-METHOD) with
    		a hack, wrapping the offending code in an opaque
    		identity function.
    		explaining CSR's insight that the ,@(WHEN ...)
    		clauses are optimizing slot access (with slots
    		encoded as FIXNUMs). Also remove DECLARE
    		NOTINLINE GET-SLOTS-OR-NIL now that I understand
    		what the code is doing.
    William Harold Newman committed Jun 11, 2002
Commits on Jun 9, 2002

    	bug reports by Alexey Dejneka and Stig Erik Sandoe
    William Harold Newman committed Jun 9, 2002

    	OpenMCL correctly flagged assem.lisp misbehavior by not
    		allowing SUBSEQ with END > LENGTH (even when LENGTH =
    		FILL-POINTER = less than the physical size of an
    		adjustable vector), so tweak VECTOR-SUBSEQ* so that
    		SBCL checks this too.
    	converting global appalling assem.lisp behavior into
    		comparatively local appalling behavior (and
    		incidentally fixing the SUBSEQ abuse)...
    	...Make rollbacks of CURRENT-INDEX and CURRENT-POSN local
    		and temporary instead of global and permanent.
    	...(also lots of minor fussing with assem.lisp: trivial
    		reformatting, renaming, correcting spelling errors...)
    	fixed undefined variable in CHECK-WRAPPER-VALIDITY as per
    		APD bug report 176 (but left the rest of the
    		reported bug unfixed)
    William Harold Newman committed Jun 9, 2002
Commits on Jun 8, 2002

    	Install slightly sanitized version of TYPE-SYSTEM-INITIALIZED (CSR
    		sbcl-devel 2002-06-07)
    	... really fix LOAD this time.
    csrhodes committed Jun 8, 2002
  2. (I was tricked by pcl-cvs into not selecting version.lisp-expr before

    I did commit. I shall strive to be so vigilant that there's no more than a
    30% chance of it happening on the next commit!)
    William Harold Newman committed Jun 8, 2002

    	workaround for openmcl 0.11 as xc host (the openmcl bug CSR
    		noted as "Extra superclass in FILE-ERROR", I presume)
    William Harold Newman committed Jun 8, 2002
Commits on Jun 7, 2002

    	Merge TONYMS patch for ignored variables sbcl-devel 2002-06-06
    	... update for new def!constant definition
    	... delete eval-stack-related stuff in non-x86 trees too
    	Fixup in ldso-stubs.S to allow building with gcc again
    	... #define LANGUAGE_ASSEMBLY (but it might be better to put this
    		in Config files eventually)
    	Document build successes on SPARC
    csrhodes committed Jun 7, 2002
  2. Fixing Alpha fixes

            Check in the _extra_ files needed for OSF/1
    	Disable PRINTNOISE in gc.c
    telent committed Jun 7, 2002
  3. Mostly Alpha fixes

    	Ported runtime to OSF/1 a.k.a Digital Unix a.k.a Tru64
            current_binding_stack_pointer and related are actual _pointers_
              i.e. 8 byte quantities.  Asm now uses ldq/stq on them not ldl/stl
            Fix bug when gc trigger called while *gc-inhibit* set: see
              new comment in interrupt.c for details
            Small amount of general comment/dead code cleanup
    telent committed Jun 7, 2002
Commits on Jun 6, 2002

    	updated <quote>weakened type checking</> stuff in
    		compiler.sgml since CSR pointed out that uncommented
    		text is unclear and out of date, and I decided that
    		commented text is not worth trying to support
    William Harold Newman committed Jun 6, 2002

    	changed --noprogrammer behavior to support people building
    		interactive cores noninteractively...
    	...renamed --noprogrammer to --disable-debugger
    William Harold Newman committed Jun 6, 2002

    	Various CLOS fixes...
    	... Fix printing of instances of classes with metaclass of
    		STRUCTURE-CLASS (thanks to Pierre Mai)
    	... ANSIfy CHANGE-CLASS (thanks to Espen Johnsen and Pierre Mai)
    	... Allow classes with metaclass of STRUCTURE-CLASS to have slots
    		again (this fix comes with a FIXME, as it wasn't a clean fix
    		at all)
    csrhodes committed Jun 6, 2002
Commits on Jun 3, 2002

    	closed bug 68: "working as designed" (since CSR points out
    		that the existence of CHANGE-CLASS makes it unsafe for
    		SXHASH to hash STANDARD-OBJECTs based on class names,
    		and there's no other obvious way for SXHASH to generate
    		nontrivial hash values for STANDARD-OBJECTs)
    	closed bug 71, since it hasn't existed for some time (DECLAIM
    		OPTIMIZE SPEED works now.)
    William Harold Newman committed Jun 3, 2002
Commits on Jun 1, 2002

    	reminding the compiler how to compile the high-level parts of
    		my Go program now that I'm not stuck in the same stupid
    		low level part any more, part II (where was
    		part I): You can never have too many IGNORE-ERRORSs.
    William Harold Newman committed Jun 1, 2002

    	( was checked in, but version.lisp-expr wasn't updated)
    	Remove apparently useless %CHECK-BOUND from transforms for
    	... useless and wrong, as they were (a) testing for the wrong thing and
    		(b) presumably being flushed.
    	... the mystery deepens, though, as this improves sbcl's performance on
    		the CASCOR and FIB benchmarks to near-parity with CMUCL.
    		Explanations welcome.
    csrhodes committed Jun 1, 2002