Evaluate removal of axis type knowledge in WCS/WCPS

[Piero Campalani]

Petascope has always been assigning an type to an axis:

petascopedb_8=# select * from ps_axistype;
 id | axistype  
----+-----------
  1 | x
  2 | y
  3 | temporal
  4 | elevation
  5 | other
  6 | t
(6 rows)

Some constraints apply to the overall set of axis types owned by a coverage:

if (previous.getLabel().equals(next.getLabel())) {
    throw new PetascopeException(ExceptionCode.InvalidMetadata, "Duplicate domain "
    + "element name encountered for coverage " + coverageName);
}
if (previous.getType().equals(AxisTypes.ELEV_AXIS) && next.getType().equals(AxisTypes.ELEV_AXIS)) {
    throw new PetascopeException(ExceptionCode.InvalidMetadata, "Domain can contain"
    + " at most one elevation axis for coverage " + coverageName);
}
if (previous.getType().equals(AxisTypes.X_AXIS) && next.getType().equals(AxisTypes.X_AXIS)) {
    throw new PetascopeException(ExceptionCode.InvalidMetadata, "Domain can contain"
    + " at most one x axis for coverage " + coverageName);
}
if (previous.getType().equals(AxisTypes.Y_AXIS) && next.getType().equals(AxisTypes.Y_AXIS)) {
    throw new PetascopeException(ExceptionCode.InvalidMetadata, "Domain can contain"
    + " at most one y axis for coverage " + coverageName);
}

Going towards n-dimensional datasets, the type of axis can blur (eg using Index CRSs), moreover for instance more than a single time dimension should be possible in the same coverage.

Completely removing the notion of axis type should be evaluated in Petascope. Alternatively the constraints should be relaxed.

Current issues are:

• index CRS (see *_ALIASES in applications/petascope/src/main/java/petascope/core/CrsDefinition.java CrsDefinition] class) axes clash with spatial axes so their type should be other and not x, y, etc.
• unique time axis constraint should be relaxed.

[Piero Campalani]

Corrige: the uniqueness constraints regard spatial XYZ axes only, not time.

[Dimitar Misev]

I thought axis type association has been removed from petascope, didn't know we still have this. Deferring to v10

[Dimitar Misev]

Not sure if it's still valid?

[Bang Pham Huu]

This ticket can be closed as the problem with contrainst axis does not occur anymore, for example: one can ingest a 5 D coverages (Lat, Long, Time, Ensemble, Forecast) and the axis type is (X, Y, t, other, other).