fix: revert np.where to xarray.where when adding vars/ constraints

[lkstrp]

#555 introduced masking and used np.where instead of xarray.where since is is "38x faster than xarray where".

But xarray is slower for a reason, since it aligns operands by dimension name, not just position. np.where requires shapes to match positionally, which is not guaranteed in add_variables or add_constraints. This PR just reverts the changes, we should add tests and release a patch tomorrow.

See https://github.com/PyPSA/linopy/pull/555/changes#diff-1a9bf239769f2af33c90e00a21f7f08e2f40d626ce06ee470bb2daa65ac06a04R788-R789

[FBumann]

@lkstrp Sorry if this causes troubles.

[FabianHofmann]

@lkstrp Sorry if this causes troubles.

no this is rather on me, I though we had the internal constraint already that masks have to have the same shape as constraints/variables

[FabianHofmann]

thanks for spotting this so quickly @lkstrp

[lkstrp]

no this is rather on me, I though we had the internal constraint already that masks have to have the same shape as constraints/variables

We could also make this explicit with a test and error message, and break things for a couple of people. Or revert and break with next minor release

@lkstrp Sorry if this causes troubles

No worries, those things can slip through. It just gets more likely with the vast amount of code produced these days, and we need to watch out

[FBumann]

Yes, PR's are getting easier those days, as you can see from mine i guess...
Im aware of the switch in effort for PR creation vs reviewing, so tell me if i can improve your review experience.

[FabianHofmann]

sorry, for the gibberish in the last comment, I was in a meeting. I would say let's revert this, in particular given the failing pypsa functions. then updating linopy to the latest and keeping pypsa v1.0.x does not break. But I would like to require mask to be properly aligned in the first place. Linopy already has so many reindexing logics that you sometimes don't know what's going on...