Juan M. Cruz-Martinez

> I don't know if I have the appetite to reopen this, but I feel the theory metadata has deviated quite a bit from the "we would like experimentalist to...

> https://github.com/NNPDF/nnpdf/pull/1678/files#diff-eb4ccea229f5af5e39ccb1cc2984780e879f823a50477da80cc65ff859717c8bR112-R115 Ah, but only on the theory part. Not sure how transactional they will be though since we need them for the old theories, unless we completely regenerate the...

I've updated E605, E866P/R and E906 to follow the new format proposed by @t7phy For the name convention I'm following what has been proposed by @cschwan, i.e., EXPERIMENT_PROCESS_ENERGY_OBSERVABLE_EXTRA stopping at...

> I do not think it would be a good to over-complicate the reader by requiring to to read so many different keys just to identify a single variable. Actually...

So I should rename: E605 -> E605_DY_38P8GEV E906 -> E906_DY_120GEV And if one wants to add another process from E605 then that will be a separate folder even if it...

Ok, I've updated now with all the suggested changes. I've left the `dataset` key since in the current naming convention it is necessary due to the following: In general the...

Ok, that will basically leave the following convention: ``` ____ ``` where `OBSERVABLENAME` is _always_ included in the `metadata.yaml` and if necessary (case of E866) left empty. The free fields...

@t7phy please have a look at the current version, if everything is ok I'll use these as the "examples" for the reader. @enocera do you want me to keep `DYE866P_old_kinematic`...

While one can reconstruct the whole dataset name from the information contained in the foldername + observable... it is impossible to know what the `foldername` is to be if the...

Also, instead of ```yaml implemented_observables: - observable_name: "ob1" ... - observable_name: "ob2" ... ``` I think it would be better: ```yaml implemented_observables: obs1: data: whatever.dat plotting: whatever obs2: .... ```...