What would you like to see added to HDMF-ZARR?
See #284 and
|
dset = parent.require_dataset(name, shape=(1,), dtype=dtype, **io_settings) |
Zarr v2 supports writing arrays with shape None (or ()). This is consistent with HDF5IO:
https://github.com/hdmf-dev/hdmf/blob/a90f8ed0588fcdd84d4af1d693e851099ced6066/src/hdmf/backends/hdf5/h5tools.py#L1278
We should do the same with ZarrIO. Doing this will remove a special handling case, and remove a deviation between HDF5IO and ZarrIO as we look to merge it into core HDMF.
Is your feature request related to a problem?
This would simplify the resolution of #277.
What solution would you like?
Use shape () and stop writing the attribute zarr_dtype=scalar but keep processing it on read in case it is present.
Do you have any interest in helping implement the feature?
Yes.
What would you like to see added to HDMF-ZARR?
See #284 and
hdmf-zarr/src/hdmf_zarr/backend.py
Line 1504 in 8ee5028
Zarr v2 supports writing arrays with shape None (or
()). This is consistent withHDF5IO:https://github.com/hdmf-dev/hdmf/blob/a90f8ed0588fcdd84d4af1d693e851099ced6066/src/hdmf/backends/hdf5/h5tools.py#L1278
We should do the same with
ZarrIO. Doing this will remove a special handling case, and remove a deviation betweenHDF5IOandZarrIOas we look to merge it into core HDMF.Is your feature request related to a problem?
This would simplify the resolution of #277.
What solution would you like?
Use shape () and stop writing the attribute
zarr_dtype=scalarbut keep processing it on read in case it is present.Do you have any interest in helping implement the feature?
Yes.