[posting problems continue.  apologies if this is a duplicate, n.]

here's some history on why you have to specify 2 or 3 coordinates
per delta in a fully regular mesh.  to be honest, i've never actually
seen it used for a real dataset, but the flexibility is there inside 
the dx data model & modules to handle a grid in which each step moved
along more than one axis, like this:

origin 0  0
delta  1  0.5
delta  0  1

this grid would be warped in a sort of diamond shape.   it's fun to play
around with.   i'd love to hear if anyone has ever used it for real data.

on the other hand, i don't think we've ever regretted being able to
specify any flattened order for how data corresponds to a regular grid.
most common are x, y, z (c order), or z, y, x (fortran order), but we've
often seen x, z, y (x-z sheets stacked along y) and all 3 remaining
possible permutations of 3D point orders.   it's sometimes been difficult
to explain the concept, but it means you never have to rewrite your large
data arrays to make them conform to some predefined point order.
that's always been a big win.

nancy

Reply via email to