Closed nickponvert closed 4 years ago
Trying to follow the thread here.
Here is the first line that fails: https://github.com/AllenInstitute/visual_behavior_analysis/blob/3a9994aa4bcf8cac6834e24e4cb6193d9edaa39f/tests/ophys/dataset/test_visual_behavior_ophys_dataset.py#L190
The ophys_data_dir
is a test fixture that we generate here, saving a bunch of stuff to h5 files:
https://github.com/AllenInstitute/visual_behavior_analysis/blob/3a9994aa4bcf8cac6834e24e4cb6193d9edaa39f/tests/ophys/dataset/test_visual_behavior_ophys_dataset.py#L125
When we initialize a VisualBehaviorOphysDataset
object, it now requires a file called roi_masks.h5
We likely need to generate the roi_masks.h5
file in the ophys_data_dir
test fixture to solve this.
Actually I think I'm thinking about this wrong. The failing test shows that this version of the VBOD object is not compatible with data that does not include roi_masks. If early data did not include that file, and we want to maintain backwards compatibility, we should tell VBOD what to do when it isn't found.
When running tests for PR #521 I found that most (if not all) of the test failures are related to a specific file that can't be found.