Currently we use the entities name for the group name in HDF5. These need to be unique. In addition with our relative flat hierarchy this leads to the problem that very often things that belong together, and would in theory have the same name must be named different. In many cases using name+type, maybe as "name@type" or "name::type" would fix that issue. Both information could still be stored as an attribute for quick access without parsing.
An example (I am sure there are better ones), an image with its data (data array), some channel information (also a data array) a corresponding ROI (also a data array), and the group that groups all these together. They would all exist in the same namespace so must be name X_image, X_roi, X_channel, and X (for the group).
Currently we use the entities name for the group name in HDF5. These need to be unique. In addition with our relative flat hierarchy this leads to the problem that very often things that belong together, and would in theory have the same name must be named different. In many cases using name+type, maybe as "name@type" or "name::type" would fix that issue. Both information could still be stored as an attribute for quick access without parsing.
An example (I am sure there are better ones), an image with its data (data array), some channel information (also a data array) a corresponding ROI (also a data array), and the group that groups all these together. They would all exist in the same namespace so must be name
X_image,X_roi,X_channel, and X (for the group).