`export_workspace` not possible when running UDP with `datacube_from_process`

[VincentVerelst]

Context

I'm trying to run a UDP from a namespace, using datacube_from_process. I'd like to save my results to a workspace using export_workspace. My UDP already contains one or more save_result nodes, so I don't need/want to add it manually anymore.

Problem

datacube_from_process returns a DataCube object. However, export_workspace only works on a SaveResult object. The following code snippet can be used to replicate the error:

cube = connection.datacube_from_process(
    process_id='worldcereal_crop_type',
    namespace=namespace,
    spatial_extent=spatial_extent,
    temporal_extent=temporal_extent,
    model_url=model_url,
)

cube = cube.export_workspace(
    workspace="worldcereal-phase2-waw4-1-workspace",
    merge="worldcereal_global_croptype_benchmark_50km",
)

will result in

AttributeError: 'DataCube' object has no attribute 'export_workspace'

Discussion

Would it be possible to allow to call export_workspace directly on a DataCube object? Or, if there's a good reason not to: how can we allow to easily add an export_workspace node to a UDP?

Workaround

For now, I can continue, by adding the following line of code between datacube_from_process and export_workspace:

from openeo.rest.result import SaveResult
cube = SaveResult(cube._pg, connection)

[soxofaan]

indeed, I noticed we encounter this problem more and more, so I started this more general ticket:

• Generic Connection.process()? #845

Would it be possible to allow to call export_workspace directly on a DataCube object?

No, that's intentionally a design that we want to avoid: the openEO process export_workspace works with a STAC resource, so it should not be a method on a DataCube object.

Or, if there's a good reason not to: how can we allow to easily add an export_workspace node to a UDP?

What you want, I think, (and that's what #845 is about), is a helper that is smart enough to give you a SaveResult (e.g. by inspecting the UDP return type), e.g.

save_result = connection.process(process_id='worldcereal_crop_type', ...)
# save_result is automatically a SaveResult object here
cube.export_workspace(...