Problem with proposed validation strategy proposed in #10

[VisLab]

This draft PR implements the proposed validation strategy of issue #10. This would result in considerable simplification, including getting rid of the HedVersion class. Since the HEDTools cache the HedSchema objects for given schema versions, once one has been created, trying to create another with load_schema_version just uses the cached version.

The fundamental flaw with this approach is illustrated in lines 67 and 68 of test_hed_tags. The HedTags object must have a HED version when it is created, but it seems that adding columns to dynamic tables does not have the flexibility of passing additional parameters.

Help would be appreciated.
@rly @oruebel

[VisLab]

@oruebel @rly would you please provide some help in resolving this issue?

[rly]

Hi @VisLab sorry for my delay in response.

This is a bug in pynwb. I created an issue here: NeurodataWithoutBorders/pynwb#1860
I'll resolve it and push a release out early next week.

In the meantime, you can do this workaround:

nwbfile.trials = TimeIntervals(name='trials', description='experimental trials')
nwbfile.trials.add_column(name="HED", hed_version="8.2.0", description="temp", col_cls=HedTags, data=[])

[VisLab]

Thanks for the response @rly. I'll try your code and will watch for the PR. Could you also remind me of how to enable the bugger for this. I didn't put it in my notes.

[rly]

We are making a new release of PyNWB in the next two weeks - sorry for the delay on that. You can use the dev branch of PyNWB if you would like to test this sooner.

To enable the debugger, comment out this line:

ndx-hed/pyproject.toml

Line 80 in 20b9f41

addopts = "--cov --cov-report html"

That line runs the test and produces coverage reports. Apparently the debugger cannot run at the same time as code coverage tracking.