[Bf-committers] Separating node systems

[Matt Ebb]

On Tuesday, February 15, 2011, Lukas Tönne <lukas.toenne at googlemail.com> wrote:

> * The design- vs. execution code separations is still a bit
> incomplete. My idea of a clean system would be that the "design-time"
> code knows as little about execution data as possible.

Yep. IMO ideally the stuff in the node editor should really be
independent of the actual processing- it's just a ui used to define a
node graph topology and set properties - then that high level info can
be used to initiate the execution process in whatever way is
appropriate for what's being processed.

Apart from a cleaner design the main reason this would be useful would
be for external renderers or other extensions to be able to define
their own node types and let users wire them up - for example creating
a shader for an external renderer. These nodes wouldn't need to do
anything for themselves in blender, it's just on a ui level - the
exporter would then just examine the graph and do it's own processing
however it sees fit. This is a pretty standard capability in other
systems like this.

Matt