base_formats: How to handle derived but incompatible file types?

[PeterKraus]

As is, base_formats is meant to handle compatible base formats, e.g. NeXus with its base format of HDF5 or the Fusion GC JSON format which is still just a JSON.

What to do for incompatible base_formats? Example from the diffrac-eva-xy discussion below:

Well, it depends. It's a format the same way generic-csv would be a format. Lines starting with # are comments, empty lines are skipped, everything else should be x.xxx y.yyy (whitespace separated) otherwise it's not a valid generic-xy.

That's kind of the problem, this eva diffract format already breaks the comment rule (which is why I had to add special support for it in datalab). I get that it might be useful though, e.g., at the very base of it it would be nice to distinguish between plain text or binary formats based on this generic format

Originally posted by @ml-evs in datatractor/yard#18 (comment)