ZarrIO needs to be updated to follow HDF5IO in how it handles:
Writing references at the end via a queue
Making sure the references are created in the same way to avoid extra checks to search for the target.
Update the workflow on write_dataset on ZarrIo, but also verify that write_dataset on HDMF is proper (specifically in how both handle references). For example, clarify why ZarrIO uses HDMFDataset as a conditional in write_dataset.
Export needs to be updates to current HDMF version
Both ZarrIO and the reference wrapper classes can be abstracted to decrease the amount of code duplication.
Currently there is an issue with convert_dtype that converts the dtype from uint32 to just int32. This can be reproduced by running the plot_io_convert_nwb_hdf5.py
Is your feature request related to a problem?
No response
What solution would you like?
Update to the backend.
Do you have any interest in helping implement the feature?
What would you like to see added to HDMF-ZARR?
ZarrIO needs to be updated to follow HDF5IO in how it handles:
uint32
to justint32
. This can be reproduced by running the plot_io_convert_nwb_hdf5.pyIs your feature request related to a problem?
No response
What solution would you like?
Update to the backend.
Do you have any interest in helping implement the feature?
Yes.
Code of Conduct