Re: [Gzz] OrthoCoordsys???

[Tuomas Lukka]

On Thu, Aug 22, 2002 at 01:33:55PM +0200, Benja Fallenstein wrote:
> >I now have made the OpenGL client boot but making it look
> >right is difficult: the new class OrthoCoordsys, which
> >is undocumented (neither the overall design nor javadocs)
> >is on the vob path and used by the VobVanishingClient indirectly. 
> >
> 
> I don't understand. What do you mean by "used indirectly?"

Through CellVobFactory --->

> OrthoCoordsys simply stores the coordinates of all coord systems in 
> arrays. Before, OrthoCoorder stored them in CS objects, which made for a 
> greater garbage collector load. But AFAIK, all there is now could be 
> implemented through the CS object approach.

No, that's not the problem. The problem is the *HIERARCHICAL*
coordinate systems, which changed a lot of APIs (and also made 
a lot of jython GL-using code not work any more).

> >Benja: could you figure out what's the quickest way to implement
> >the similar stuff for OpenGL or alternatively, get rid of it?
> 
> What is the problem? Doesn't the GL stuff have a way to store 
> coordinates already?

The hierarchy.

> >I'll do the implementation once you document and explain.
> >Note that the resulting code needs to be FAST: if it isn't fast enough,
> >we need to scrap this approach.
> 
> I don't see how something that makes storage in the AWT vob system 
> faster could influence the speed of OpenGL at all, much less make it slower.

Again, the hierarchy.

        Tuomas