Closed rmndrs89 closed 1 year ago
Thanks, Robbin for the critical feedback. This is valuable, especially when writing examples for people to import their own data.
I should add a function to the postinit process that derives the types
and ch_names
automatically. I included these fields, as they are close to the MNE implementation. For some future functions, I see some advantages to having them easy to access. Let's discuss this in our next meeting
I updates the dataclasses for now, also made adjustments to the importers in https://github.com/neurogeriatricskiel/NGMT/commit/10af5ec4ce0269324d165d3cc50c253680e55689. Please check if it makes sense. I also updated the documentation a little. Let's still discuss tomorrow, if we are happy moving forward with the current implementation. For me, ch_types
or ch_names
should be a MUST at some point, but if provided once can be fetched from any class to avoid redundancy.
Hey guys,
I am still having troubles understanding our complex dataclass structures. Now I have adopted the importer for the Keep Control dataset, but when I was testing it, I run into an issue related to: https://github.com/neurogeriatricskiel/NGMT/blob/28c6d94c6dc4203ba8a15795dff465994a26d31b/ngmt/utils/ngmt_data_classes.py#L143
Why is there a function that checks https://github.com/neurogeriatricskiel/NGMT/blob/28c6d94c6dc4203ba8a15795dff465994a26d31b/ngmt/utils/ngmt_data_classes.py#L152?
The error I got is:
That means that if I do not specify
ch_names
for the RecordingData, then it is of typeNoneType
and the comparison cannot be made.Should we remove
types
andch_names
as data attributes, and simply derive them from theChannelData
attribute (these should contain the same information anyway)?