Andrew Tritt
Andrew Tritt
Yes, I think this is no longer an issue. @rly can you confirm?
Yeah, I would say change it so they're all ____fields____
Can you clarify what you mean by “sparsely”?
Yeah, you can just pass `fillvalue=0` through the same mechanism I suggested [here](https://github.com/NeurodataWithoutBorders/pynwb/issues/1067#issuecomment-537114178) and you implemented [here](https://github.com/hdmf-dev/hdmf/pull/159#issuecomment-537154952)
Ah, I see what you're asking. You would need to use an [AbstractDataChunkIterator](https://hdmf.readthedocs.io/en/latest/hdmf.data_utils.html#hdmf.data_utils.AbstractDataChunkIterator) to return just the chunks you want to write. My understanding that HDF5 only writes the fillvalue...
@bendichter @oruebel Where are we on this?
At this point, I think the only solution to getting the behavior that you are expecting would be for us to inspect the data you pass in under the hood,...
> I think ultimately though it is a bug for a list of strings to be automatically turned into a string of that list. @bendichter Yeah, this lies somewhere in...
@tjd2002 > Would this allow two or more spikes from different units with the same timestamp? (This can happen in real data) Yes, there are no restrictions on unique time...
I am in favor of NWBSession or NWBExperiment. PyNWB has a classed called 'NWBDataset', and there's a primitive concept of dataset in FORM and HDF5.