Open hjabbot opened 5 months ago
The working plan is to create a suite of unit tests for the abstract dataloaders, and then it is up to the end user to make sure that each individual dataloader is correctly formatting the data to be handled by the abstract dataloaders.
A separate suite of unit tests can be saved in a private repo that will check the header info of each data source to flag any changes that occur. I don't think that anything beyond the header info is worth testing, as data changes and so there is no 'ground truth' that the data itself can be compared against.
Moving here from project repo #151
The GRF and CSV dataloaders can get this implemented no worries, and realistically they should be done ASAP as our regression tests are dependant on the GRF dataloaders working consistently.
Need to figure out issues with EULA's of the data products we're ingesting first. Perhaps we could create fake datasets and attach real header information just to see if the functionality still works.