Open mktippett opened 2 weeks ago
Thanks for the report and the example. This is really crazy, the slow example takes 2min on my machine while the fast one is basically instant.
After digging a bit, the problem seems to be, that xr.plot.dataset_plot._temp_dataarray
broadcasts everything against everything creating a super large array of shape (12, 12, 250, 7, 30)...
@Illviljan do you have any idea? Probably removing unessesary coords before the broadcast might help?
What happened?
scatter plot is slow when the dataset has large (length) coordinates even though those coordinates are not involved in the scatter plot.
What did you expect to happen?
scatter plot speed does not depend on coordinates that are not involved in the scatter plot, which was the case at some point in the past
Minimal Complete Verifiable Example
MVCE confirmation
Relevant log output
No response
Anything else we need to know?
For me, slow = 25 seconds and fast = instantaneous
Environment