Data Structure and manipulations

[zeffii]

It occurs to me that we

• don't often create triangle based geometry, and operate on it with lists even less often.
• It's far more common to create quads.
• It is very uncommon to need irregular structures like [[[0,1,2],[3,6,7,8,12,13],[4,2,5]]],
  • except for final polygon representations,
  • or in particular sliced geometry (keeping track of the edges, vertices associated with slices).

It may take a while to get used to the idea that the First class storage of our data should be flat or homogeneous arrays, and that any specialized notions we have about grouping data should be possible to store in lookups which themselves are also flat or homogeneous 2d lists. Any operations on this specialized lookup data may be tricky to comprehend at first, but it's just data -- we will figure it out.

[nortikin]

if we import from scene it could be anything and three and four and more

[nortikin]

it become more clear when we know flat data will be unchanged. but if we change values it will be changed, not order of flat

[zeffii]

We already know how to efficiently store irregular content like mixed tri/quad/ngons, I described it in np_Mesh. What will be interesting is how to perform the operations on the lookup tables :)

I understand why you are concerned, but we already solved (in the past) more difficult things. imo

[zeffii]

If in the past we solved a problem one way, then now we can approach it from a different perspective. If we can not nicely solve it another way, then we can solve it using the original method but the speed will be the same as the old way.

We lose nothing by trying to rethink how this is approached, the redux isn't intended to be a 'restatement of original code but prettier' it is ' find those parts of our animal which can benefit from being rethought and apply numpy to those'

It would benefit us if we describe problems by their intended outcome rather than the datastructure previously used. Images. examples. cases. Think of the problems as being solved by a black box at the moment.

[zeffii]

http://technicaldiscovery.blogspot.nl/2010/11/zen-of-numpy.html

[nortikin]

there are some terms i need to study. Study Zenumpy.