Update load_motion_info to load legacy motion folders
#3772
Conversation
There was a problem hiding this comment.
I know Sam prefers bool check over doing try/except (at least he seems to). What if we flipped the logic and checked for the legacy files and if all legacy files are present we load legacy. If not we load the current format and let the current format raise the IOError if its neither type. try/except are expensive if the except is used. This is probably the simplest version, but just wanted to raise the point.
Otherwise this looks fine to me.
EDIT: error appears to be related to building scipy. Not this PR.
|
Good point @zm711 Got rid of try-except :) |
|
Looks good to me. You want to do another PR to figure out the scipy build issue? And do we have any way to test? Or do we not have a way to generate (simulated) legacy motion? |
|
Let's fix scipy in another PR. I don't think we need to test this since we would make up those three files... Here it is in action: |
| temporal_bins_s = [np.load(folder / "temporal_bins.npy")] | ||
| displacement = [np.load(folder / "motion.npy")] | ||
|
|
||
| motion = Motion(displacement=displacement, temporal_bins_s=temporal_bins_s, spatial_bins_um=spatial_bins_um) |
There was a problem hiding this comment.
| motion = Motion(displacement=displacement, temporal_bins_s=temporal_bins_s, spatial_bins_um=spatial_bins_um) | |
| motion = Motion(displacement=displacement, temporal_bins_s=[temporal_bins_s], spatial_bins_um=[spatial_bins_um]) |
There was a problem hiding this comment.
He already made them lists above. Do they have to be nested lists?
There was a problem hiding this comment.
No :) I'm ok making the list later if it's better readable
There was a problem hiding this comment.
Actually, let's leave it as is!
When we added the
Motion.loadto theload_motion_info, we forgot to handle back-compatibility.This PR just tries to load a legacy motion in case the
Motion.loadfails.