[EMAIL PROTECTED] wrote:
>
> Hi all,
>
> [EMAIL PROTECTED] wrote:
> >> [EMAIL PROTECTED] wrote:
> >> >OK - found the problem, I think. Similar to the original problem where
> >> >there are two sources of culling information, but for now for normals
> >> >rather than post-clip vertices. I'm committing a fix as I type this.
> >> This one seems to fix one of the two errors. The lament mode of xlockmore
> >> does not crash anymore.
> >> The other error (I labeled it error 1 in a previous mail) in the stairs mode
> >> is still there. Also for this problem holds : it first occured when the kw3
> >> patch was merged into the CVS-tree.
> >>
> >> Jouk
> >>
> >
> >I've tried but can't reproduce this problem. Do you feel confidient to
> >expand the macro definitions of persp-tri-xyz in order to get the line
> >number where the problem occurs?
> I think you are right Keith. I found the problem. It seems to be a bug
> in my compiler (DEC C V5.7-004 for VMS) If I switch optimization off when
> compiling triangle.c the problem seems not to occur (grmpfff&^%()
>
> I will have to report this to DEC.
> Question: Is it alowed for this purpose to give DEC a snapshot extracted from
> CVS-tree?
Ahah - a bug that's not my fault. I actually think we should preserve
it in th sources forever - like some kind of incredibly rare butterfly,
pay it the respect it deserves....
I can't see why we wouldn't be able to give HP (or anyone else) a
snapshot.
Kwirh
