Closed adamltyson closed 3 months ago
Be ruthless and remove any feature using third party tools we can't rely on?
Not sure either - I'd rather not ignore tests for 3.11 if possible, and we've put in a lot of effort to be able to move to 3.11.
Is contributing to the allensdk
an option? Otherwise, my vote is probably this :arrow_up:
Sorry, my mistake. Just FYI it's 3.12 causing the problem.
We may just have to bite the bullet and remove AllenSDK from all of BrainGlobe. It works well in it's own environment, but it doesn't seem to be maintained sufficiently to rely on as a dependency. Maybe we can hold off to see if https://github.com/AllenInstitute/AllenSDK/issues/2744 is addressed, otherwise we can go through and remove features that rely on the SDK?
I've updated this PR to only remove 3.9 (not add 3.12).
I've raised https://github.com/brainglobe/morphapi/issues/67 and https://github.com/brainglobe/morphapi/issues/68 to track the issues raised here.
Tests are failing due to this issue. Unless the Allen can address this quickly, I'm not sure what the solution is. Ignore these tests? Be ruthless and remove any feature using third party tools we can't rely on? Hold off updating to 3.11 for a while?
Although 3.12 is the issue here, the AllenSDK pins scipy, xarray, pandas and numpy to fairly old versions which may cause issues when installing the metapackage.