Open h-mayorquin opened 2 years ago
I'd vote for frames
Although, specific to the usage in frame_to_time
, how badly do we want to mimic SpikeInterface and plan to eventually just have a single get_times
function that either uses an underlying irregular time_vector
or otherwise returns a np.arange
scaled by the sampling frequency and shifted by an offset?
Yeah, maybe going for get_times
for consistency is better overall.
Right now we have
frames
on the functions for getting time: https://github.com/catalystneuro/roiextractors/blob/master/src/roiextractors/imagingextractor.py#L69And
frame_idxs
in the function for getting the frames: https://github.com/catalystneuro/roiextractors/blob/b71f17db4d3dadeaa4ae02c806c1e269f03d3d9b/src/roiextractors/imagingextractor.py#L62-L68 But both of them are referring to the same. We should use eitherframes
,frame_idxs
orframe_indices
for the two of them. Numpy usesindices
for their indexing routines so maybe that's a good choice: https://numpy.org/doc/stable/reference/arrays.indexing.html#routines-indexingBut I don't think it matters that much.