Is your feature request related to a problem? Please describe.
Currently, the only way to tell the script what area a given dataset covers is to specify a rectangular bounding box as the bbox element of the source definition. But many useful datasets do not fit this constraint! For example, the data source used for https://github.com/eldang/elevation_lookups/pull/31#discussion_r1014166244 is limited to the UK, but the SE corner of the bounding box is over France. And the data source in https://github.com/eldang/elevation_lookups/issues/32 is shaped like this:
The practical effect of this is that in the corners which have no data, the script has no way of knowing that it should fall back to a data source further down the list.
Describe the solution you'd like
I would like to address this by adding a boundary element, and making the script use that if present instead of bbox to define the valid extent for a data source.
Describe alternatives you've considered
So far we've worked around this by using bbox approximations, and it's been OK because we haven't had overlapping data sources. But if we start to fill in more areas it will become a real problem.
Is your feature request related to a problem? Please describe. Currently, the only way to tell the script what area a given dataset covers is to specify a rectangular bounding box as the
bboxelement of the source definition. But many useful datasets do not fit this constraint! For example, the data source used for https://github.com/eldang/elevation_lookups/pull/31#discussion_r1014166244 is limited to the UK, but the SE corner of the bounding box is over France. And the data source in https://github.com/eldang/elevation_lookups/issues/32 is shaped like this:The practical effect of this is that in the corners which have no data, the script has no way of knowing that it should fall back to a data source further down the list.
Describe the solution you'd like I would like to address this by adding a
boundaryelement, and making the script use that if present instead ofbboxto define the valid extent for a data source.Describe alternatives you've considered So far we've worked around this by using
bboxapproximations, and it's been OK because we haven't had overlapping data sources. But if we start to fill in more areas it will become a real problem.