Re: [Fresco-devel] Bogus traversals (and patch)

hi Tobias,

[ please don't send mails to the list *and* to me.]

Tobias Hunger wrote:

> This sounds like a clean solution to me. Except for one thing: Have one method 
> 'continue_traversal'  for both,  MonoGraphic and PolyGraphic. That way a 
> custom Traversal does not need to figure out what it is dealing with before 
> being able to continue.

huh ? The traversal doesn't need to figure out anything. It's the 
graphic itself that calls 'this->traverse_body()'. But I agree, we don't
need two separate methods. I guess all I wanted to make clear is that 
this method doesn't need to be exposed through the Graphic (CORBA) 
interface, it can be part of 'MonoGraphic' and 'PolyGraphic'.

Read: we can well stipulate that we use the same name, but they are two
separate entries in two distinct vtables.

> The downside is that we will need to expose yet another method via CORBA. 
> Without that such a client-side traversal Nicholas has in mind  is not 
> feastable. Nicholas' proposal did not result in the need to change the 
> interfaces, it even makes it a little bit harder for client-side graphics to 
> break the rendering alltogether.

I hate to say this again, but wouldn't it be better to wait with these 
discussions a bit ? We want to discuss some architectural changes to the
scene graph (for example not exposing traversals through CORBA any more) 
so this has a big impact on our current arguments, doesn't it ?

Of course, this whole thing is helt back by the (not really) forthcoming
release. Any help to get done with that would really be apreciated. I'm
currently trying to finish the website...:(

Regards,
		Stefan