Closed clbarnes closed 1 year ago
Test failure seems unrelated.
At least some of the test failures are from the to_scipy_sparse_(matrix|array)
fix in navis, which isn't released yet.
The remaining failure seems to be from the server used for unit tests returning a 500. Ideally these responses would be mocked rather than reaching out to a server (would save server load and also speed things up).
Hi Chris. Not sure if you were waiting for me to chip in here. In case you were: looks good to me.
Also thanks for the reminder to release a new navis
version
The original implementation for fetching the annotation graph ignored volumes, and made some assumptions which fit a particular use case I've now outgrown :sweat_smile:
I've deprecated it in favour of the new
get_entity_graph
. It's more generic over semantic objects, but there's still an escape hatch for getting skeleton IDs rather than neuron IDs (the difference is documented). You can key entities in the graph on name or ID; the caveats of using name are documented and the errors should be informative. You can also cut down the result by annotations, which will hopefully speed it up.Some last-minute additions still need some more testing.