[Bf-taskforce25] data api proposal

[Campbell Barton]

On Tue, Sep 30, 2008 at 10:45 AM, Michael Fox <mfoxdogg at gmail.com> wrote:
> hi all just a few thoughts that are jumping out at me
>
> - all this time i have been thinking we used shorts for bit flags as its
> memory efficient, but changing everything to int and floats wont that
> make blender's memory usage especially file size become huge as ints use
> many bytes same with floats.
don't think we should change this now, probably true for many data
types but for faces, edges, verts, bez-triples its still good to store
selection, modes etc in char/shorts, and we will have UI/property
access to some of these.

> - the only limits i have seen has been when you are defining buttons, i
> agree there should be a scale of some sort eg, have soft limits like
> 0.5,1,5,10 etc ect, but how would the user know there is a huge limit or
> small limit, i suggest automatically be generated in the tooltips like
> documentation as you define the limits when you define the access
We'll definitely need a good way for users to go above soft limits
with UI widgets, lightwave just let you type in bigger numbers then
the slider limits, I hope we can do better.

> - adding  new layer on top of dna would really hamper development if
> there is no optional direct access to just quickly hack together
> something or you had an idea and just want to add another variable to
> the struct, you don't want to define a dozen more functions before you
> can even access it or even see it as it will cause alot of mess, but for
> a patch to be accepted proper access must be made.

This is not intended for blenders internal functions which can use
struct members directly. only UI, PyAPI, Ipos, proxies, renderAPI? -
so it wont stop you from writing hacks :)