[Soc-2007-dev] Friday Status Report

Aaron Moore two.a.ron at gmail.com
Sat Jul 14 02:11:49 CEST 2007


Hi everyone,

I had a fairly productive week, here's what I did:
  1. wrote the object type sorting functionality of get_objects, and
get_instanced_objects
  2. wrote the instance sorting functionality for get_instanced_objects
  3. wrote a preliminary version of the dupli-objects collapsing functionality.
  4. made get functions for: lights, geometry (only polygon meshes so
far), vertices, edges, faces, indices.
  5. made the iteration functions of all of the above.
  6. got plugins loading. Lots needs to be done before this is fully
functional, but the preliminary step is done. I also need to
coordinate with Peter about furthering the development of his plugin
system, which I will need to help do.

What I plan to do next week:
  1. code the rest of the preliminary api.
  2. create a test RIB exporter plugin.
  3. continue the design of the Render API interface.
     - add a few needed functions to the basic api, mostly concerning animation.
     - design functions for other types of geometry

This is not very complicated coding, it's basically a matter of
tracking down the right fields in structs or functions in the header
files that I need to use. One concern I'm having is that, for example,
the DerivedMesh functions may not maintain the vertex order of the
geometry. I don't see why it would do this, but since blender doesn't
explicitly need to know that two versions of a deforming mesh are the
same (because of the way it does motion blur) it might not be coded in
such a way that the requirements of the render api are met. This is
one instance of a general problem of having functions that derive data
I need in blender that don't quite do what I need them to. In these
cases I would have a dilemma:
   - code custom functions for the render api, or
   - alter the blender functions so that they conform to the
requirements of the render api.

The first would create duplicate versions of the same functionality:
never a good thing. The second would make blender share the render
API's requirements.

In any case, I haven't found this problem yet, because I haven't
gotten to the places that might be problematic in development yet.

Aaron


More information about the Soc-2007-dev mailing list