Adding default coordinates to Datasets with missing image coords#605
Conversation
|
Some datashader related tests seem to be failing on Python 3. Python 2 is passing because:
|
|
The datashader test failures are not being caused by this PR so that is a separate issue to fix. Restoring the original commit. |
|
Amazing, thanks for your work on this! I would make sure to test many possible combinations here (no coordinates, some coordinates, dask, no dask, different ndims, etc.) The example I reported in #603 was just one fail case. Every different scenario seemed to fail with a slightly different error message, and unfortunately I did not take the time to document all of them. |
| elif self.datatype == 'xarray': | ||
| import xarray as xr | ||
| if isinstance(data, xr.Dataset): | ||
| if kind == 'image': |
There was a problem hiding this comment.
Yup, all need to be handled. PR is still WIP and not ready for review. The first commit was simply tackling the particular example reported in the issue.
Co-authored-by: Philipp Rudiger <prudiger@anaconda.com>
Absolutely! I'll make sure there are unit tests for these cases. |
jlstevens
force-pushed
the
jlstevens/default_image_coords
branch
from
ecbd91c to
fc500d3
Compare
|
Going to merge this as a first cut, please do follow up on all the other cases but I'm releasing 0.7.2 tonight so wanted to get this in. |
3 participants
This PR aims to address #603 by generating default image coordinates when they are missing.
Before the PR, using the example given in #603:
With the PR:
So far I've only thought about the
Datasetbranch, to finish this PR I'll need to handle theDataArraycase as well.