You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
is there a reason one would not want to always include all coordinates in the loadable_variables
The main reason though is that coordinates can be N-dimensional in general, in which case you might use a lot of storage duplicating them. But for low-dimensional coordinates I agree we should recommend people load them.
Perhaps we should even add another option: loadable_variables='coords' or loadable_variables='1d_coords', which we could default to... The latter is pretty much what xr.open_dataset already does for choosing which coordinates to create indexes for.
In fact if we do this then why not just change the default for indexes too, or even get rid of it entirely. Loading 1D coordinates and creating indexes, but keeping 2+D coordinates virtual and not trying to create indexes for them (which I don't think is even possible in xarray right now anyway) would be much more intuitive and sane default behaviour, and it would mean all our examples don't need the indexes={} kwarg in them anymore.
In other words we want the default behaviour to be the same as xarray except that data variables and multi-dimensional coordinate variables are virtual. I think I was just over-complicating things.
In #353 I added the options, but making them work with the indexes kwarg for all possible cases is a lot of effort for basically no gain as far as I can tell. I think I will just remove indexes entirely.
Originally posted by @TomNicholas in #330
The text was updated successfully, but these errors were encountered: