Closed willirath closed 4 years ago
Does xarray already do this as a loading arg, or are you asking whether intake-xarray can (optionally) perform the rename
?
I don't have a huge objection to this, particularly since it is a no-cost operation.
Having said that, Intake has tried to shy away from being a place where data processing happens, and it would be good in general to try to keep to a minimum the number of extra arguments supported by the intake driver.
Does xarray already do this as a loading arg, or are you asking whether intake-xarray can (optionally) perform the rename?
I'm asking wether the latter (performing rename after the loading) is within the scope of intake-xarray. But I fully understand your approach of keeping this as simple as possible.
I guess the solution then is to think about how to do essential post processing in the cleanest way. As this is more of a usage question, I'll move over to stackoverflow.
For datasets that do not follow good naming practices, I'd love to be able to just
cat["entry"].to_dask()
and get a clean dataset.While all the renaming necessary for this could be done with
preprocess
via xarray keyword args, I wonder if there's a way of just specifying a renaming dict in the catalog.