Change quantitative properties to add units to name

[alessandratrapani]

Change quantitative properties from attributes to datasets and add unit property as an attribute of each dataset that is numerical. Additional adjustments:

• [x] for datasets: change required:false --> quantity:'?'
• [x] remove unnecessary class definition from patterned_ogen.py
• [x] unit property would not be modifiable
• [x] add dims and shape
• [x] add unit measure in the name of the attribute (https://github.com/NeurodataWithoutBorders/nwb-schema/issues/569), instead as defined them a (scalar) dataset + unit-attribute

[CodyCBakerPhD]

Some conflicts here now

[CodyCBakerPhD]

The main reasoning for this was to be able to set a unit attribute then on each quantitative dataset, right?

Would you be able to share an example file using this alteration to see how it looks in practice?

[alessandratrapani]

The main reasoning for this was to be able to set a unit attribute then on each quantitative dataset, right?

Yes, sorry it's still work in progress. I will have a meeting with Ryan tomorrow to solve this

[alessandratrapani]

Here is an example file. Not quite sure why tests are failing here, but not locally.

[CodyCBakerPhD]

Not quite sure why tests are failing here, but not locally.

Did you try a fresh environment that is set up in exactly the same way the CI does? We as developers can often do little things differently on our own systems

[CodyCBakerPhD]

It's weird; the shared file has an excitation_lambda field as a part of its PatternedOptogeneticStimulusSite but I don't see that field specified in the schema? Any ideas (is it purely inherited)? I just notice because it doesn't have a unit like the others do

[CodyCBakerPhD]

@alessandratrapani Thanks a bunch for working on this, it's great to see an example in practice

Just something to leave as a note here; adding lots of tiny datasets to the file may affect standard performance of read operations, especially from the cloud. Which might diminish some of the gains we've made with the latest version of the extension overall

I'm banking on us figuring out better ways to read from the cloud (kerchunk/consolidated metadata) to resolve that, but good to be aware of and consider if the gain in specificity of the metadata values is worth it

[alessandratrapani]

It's weird; the shared file has an excitation_lambda field as a part of its PatternedOptogeneticStimulusSite but I don't see that field specified in the schema? Any ideas (is it purely inherited)? I just notice because it doesn't have a unit like the others do

Yes, it's purely inherited.

[CodyCBakerPhD]

@alessandratrapani Is there a stub file with the new attribute name change approach instead of datasets?

[alessandratrapani]

@alessandratrapani Is there a stub file with the new attribute name change approach instead of datasets?

yes, sorry I forgot to update it: here

[CodyCBakerPhD]

I think this looks great this way