Re: [Dri-devel] VB and IM

[Keith W. <ke...@va...>]

Andrew Richardson wrote:
> 
> On Tuesday 30 January 2001 02:53, you wrote:
> > Andrew Richardson wrote:
> > > Dear all,
> > >
> > >   Im hacking around inside MESA and was wondering firstly what the
> > > relation ship between the VB and IM is. VB gets filled then emptied when
> > > full and is an array, IM is one shot? Also VB->Start gives a handle on
> > > which verticies are transformed, from ONE end of a loop. Which variable
> > > is then at the other end of the loop. for(int i=VB->Start;i< ??what??
> > > ;i++){} : even implicitly this is what seems to be happening, right?
> >
> > It sounds like you are looking at 3.4?
> >
> > An IM is a struct filled only via calls like 'glVertex', and consists of a
> > bunch of fixed size arrays.  These are filled in such a way that no
> > information is lost from the original calls, so that the IM structs can be
> > stored in displaylists and replayed later.  They are also filled in a very
> > efficient manner, so that they can be used as intermediate storage in
> > normal execution of begin/end objects.
> >
> > A VB is the intermediate storage and state during the transformation of
> > vertices.  This data can come from either IM structs (normal execution and
> > the replay of display lists) or from vertex arrays.
> >
> 
> So EVERYTHING passes through VB at one time or another?

Yes, but bear in mind that 3.4 has 2 vb's, one specially for CVA
transformation.  In 3.5, there is only one that does everything.

Keith