Crosscut

Normalizing record types is pretty straight-forward. There's a bit more to variant types though. I can think of at least three aspects that need to be considered when normalizing those.

The first one is similar to what's required for record types: Sorting the cases of the variant, for example alphabetically. Because if a function expects a variant { A, B }, and what you have is a variant { B, A }, there's really no reason why that shouldn't just work; same way it would for records.

Tomorrow, let's continue with more aspects of normalization that are exclusive to variant types.