Additional endpoints to get more details about the data cube?

[m-mohr]

Inspired from D34:

Meaningful metadata and collection naming: The users mentioned that meaningful metadata is crucial to effectively use openEO. > Especially the availability of data within the extents of a (irregular) time series has been pointed out as important information.

Would it make sense to get details on data cubes easily using a separate endpoints (e.g. GET /collections/:id/datacube), e.g. specifying values/extents for all/a subset of the dimensions and then get a full list of available dimension labels etc. Especially important for temporal I think, as the general metadata usually just lists the temporal extents, not distinct timestamps.

A different approach would be to use load_collection and debug...

Another comment from D34:

Several concerns are raised about data collections and backends. An interactive exploration of data collections and their availability across all back-ends is missing, which is a barrier for users.

[soxofaan]

another approach that would not need new API endpoint:

load_collection without temporal/spatial extent
dimension_labels(dimension="t")    result=True

pro:

• does not need new API endpoint, just use standard processes
• could also work on processed collections (e.g aggregate_temporal)

contra:

• it doesn't look like collection metadata query
• backend needs to be smart enough to just use metadata and avoid loading all the actual collection data

(related to Open-EO/openeo-processes#195)

[m-mohr]

That is basically the equivalent to "A different approach would be to use load_collection and debug...", but use dimension_labels instead of debug.

Additional contra:

• I would expect that data discovery doesn't incur any costs, but sending it as a job would somehow imply that
• You can't do it without credentials, which is making discovery harder than it needs to be