embedding metadata in openminds python repository vs importing from remote source

[PolarBean]

I noticed that in the definitions for the different classes there is also all of the metadata instances which has been collected into openminds? For instance the actual code to define the BrainAtlas class is only 185 lines long. But then the file is ~5500 lines long because it contains each instance of a BrainAtlas which has been indexed in openminds. Would it maybe be cleaner to have the metadata class definition seperate from the metadata instances themselves? given that this data is also hosted here https://github.com/openMetadataInitiative/openMINDS_instances would it be possible for the python package to fetch the data from here on the user side after installation? Apologies if I've misunderstood anything!

[apdavison]

We decided to embed the instances so that

• the library could be used without needing internet access
• importing the library is fast (since no need to fetch any data over the network)

We did consider packaging the instances as a separate Python package, and if the size of the instance library gets too big we will do this, but at the moment the entire source package is only 1.1 MB.