Re: [OctDev] Java/OpenGL-based graphics package for octave

[Daniel J Sebald]

Michael Goffioul wrote:
> On 4/24/07, John W. Eaton <address@hidden> wrote:

> I don't think you have to view it that way. A hggroup object is just
> the grouping
> of other objects. When rendering a hggroup, you just have to recursively
> render the children. The actual type of the hggroup is not relevant when
> rendering, because the operation is passed to the children. The only things
> that you need to render are the core objects; all others are just 
> combination
> of these core objects.

This is what I was afraid of.  "just" implies more work on Octave's part.  That 
is what I meant the other day by primitives.


> In Matlab, a stemseries object has 2 children:
> - a line object for all stems: the key here is that the original x/y
> data are
> converted into [x0 x0 NaN x1 x1 NaN...] [B y0 NaN B y1 NaN...] where B is
> the base value; line segments containing NaN are not drawn by Matlab
> - a line object for the markers
>
> The base line is a (extended) line object that is a child of the axes and
> shared by all stemseries objects within the same plot.

Oh, that's interesting.  So the stem series is just a single line in some sense 
with a bunch of blank spots along the way.  Now, I think gnuplot can do that, or 
it is eventually intended to properly deal with that in the near future if we 
could have agreed on things.  We've talked about consistency in gnuplot's 
behavior for handles NaN vs., say, "missing", etc.

Dan