Change theory database from .db to .yaml

Created by: scarlehoff

We have many people working on the database at the same time, which is creating constant problems because it is a binary file... maybe we should move all the information to a .yaml file with id: information? I don't think we lose much.

assigned to @enocera

assigned to @enocera

@scarlehoff I definitely agree. We should get rid of theory.db. Couldn't we just use this https://github.com/NNPDF/theories/tree/main/data/theory_cards? And remove any dependencies on theory.db in validphys? Or are you suggesting that we need another .yaml file on top of these in which we have a description of the various theories?

Created by: Zaharid

I think that dumping those theory cards in the nnpdf repo (and changing the docs and the various scripts that read the db) should essentially do the trick. It does involve some work. Also we, would want to have a schema that is as good as the sqlite one (validobj should give us valid theory objects).

FWIW this was one of those things that would happen in Milan many years ago. I am sure it was in one of these blackboards that mysteriously dissipated.

Created by: scarlehoff

FWIW this was one of those things that would happen in Milan many years ago.

That is basically the cards there https://github.com/NNPDF/theories/tree/main/data/theory_cards it did happen.

Created by: Zaharid

While we are taking, we should think of a way to specify how a thoery relates to another (which sql is good for but oh well). I mean things like scale_variation_for, alpha_s variation for and so on. So we don't have to have some hardcoded piece in an unrelated part, or worse, repeating the information in every runcard and getting it wrong.

That is basically the cards there https://github.com/NNPDF/theories/tree/main/data/theory_cards it did happen.

except for

I think that dumping those theory cards in the nnpdf repo (and changing the docs and the various scripts that read the db) should essentially do the trick. It does involve some work. Also we, would want to have a schema that is as good as the sqlite one (validobj should give us valid theory objects).

Created by: RoyStegeman

How about a simple .csv instead? It may be easier to find a nice reader with filter functionality for that (e.g. any excel-like program will do) than for yaml.

mentioned in issue #1756 (closed)

closed