Closed nevrome closed 1 year ago
I like the idea, in principle. But how would this break existing workflows? Right now, when a user calls trident validate
they can trust that it tells them whatever is wrong. In the future this would then i) not anymore work, and b) ii) make them require to enter quite a long command line to replicate the old behavior.
Perhaps we should add an option trident validate --full
or something, which we could even recommend if users try to run trident validate
in a vanilla way?
Nono - there is a misunderstanding here: trident validate -d PATH
should behave as before, so crawl all directories under PATH, find all packages and validate them completely. But for example trident validate --pyml PATH
should only validate a single POSEIDON.yml file.
Sounds good then!
This would be useful in day-to-day work and for automatic pipelines where only individual files matter, e.g. the .ssf file in poseidon-eager. I imagine a simple API like this:
This allows to parse/validate a package collection (as before), or an individual set of genotype data files, or an individual POSEIDON.yml, .janno, .ssf or .bib file. We could also make it a bit more flexible by, e.g. allowing multiple files of one type or even multiple files of different types. I don't know how flexible this has to be. I assume most of the time it will be applied to one file only.