WCPS does not distinguish between coverage and CRS axis

[Piero Campalani]

This ticket addresses the same issue of ticket #506, but at a WCPS design level.

Petascope WCPS has a 1:1 binding between a CRS axis subset in a query and a domain element of a coverage, and as well a 1:1 domain and cell-domain elements of a coverage.

This means that dimension boundaries are expressed via scalar values, which are then meant to compose the N-D coordinate tuples (where N is the dimensionality of the coverage and of the CRS). This prevents a smooth implementation of future rotated/oblique/non-uniform/curvilinear geometries: for a concrete example, in the simplest case of an oblique line in a 2D CS (1D gridded coverage) going from point A to point B (4 points):

     Lat

      A
      |
  hi2 |------------x B
      |           /:
      |          x :
      |         /  :
      |        x   :
      |       /    :
  lo2 |------x A   :
      |      :     :
      o------+-----+-------->  Long
            lo1   hi1

In this case we would need a vector representation of the domain extent, so that lo and hi are {(lo1,lo2),(hi1,hi2)}, so they are not DimensionPointElements anymore. Otherwise we could keep scalar notation but the dimension is the CRS one, so we loose the mapping from the grid axes to the domain axes.

[Dimitar Misev]

Closing as out of date with respect to the latest rasdaman version.