[Bf-python] memo + new MetaBall

[Michel Selten]

Hi,

On Wed, 2003-05-21 at 03:07, guignot wrote:
> (for vectors or coordinates, we can use (Python point of view) set(list) 
> or set(val1,val2,val3). Or both. I choosed the latter but it can be 
> easily changed.

This is exactly the same as in Object with location (x,y,z), rotation,
delta location, .... functions. I'm pretty sure that I've currently
implemented it wrong :)
At least there's a reference (blender 2.27) for me to look at. I think
your set functions should have the same effect as in the Object module.
So it wouldn't be to hard to do a quick check in 2.27.

You brought up a very good point about homogeneity! Perfect.

> Access to compound data :
<snip>

Please look at how the IPO module does this. It has curves as compound
data. You'll probably need to write a sub-module for the Metaballs to
work and keep the reference to those sub-modules in the 'mother'
Metaball. Another example is how Willian implemented the vector as a
compound module in NMesh.

> If you feel concerned by these issues, please answer asap, the longer we 
> wait, the more code will have to be modified to follow the guidelines...

That's why I asked you to first write down the api for the modules
you're implementing. You're doing good work, but you've skipped one
important step :)

It would be really bad if the modules you implement for 2.28 need to be
changed in the API already for blender 2.29. An API of a module is
something that needs to be stable for a very long time - that has been a
problem with blender in the past. I hope we don't make the same mistake.

The reason why IanWill and I haven't written any api documentation yet
is because we are 'simply' trying to implement what's already known in
2.23 and 2.25. If we'd be implementing new modules, we first have to
write down how the api for such a module would look like.

With regards,
	Michel