[esgvoc] Institution drs_name inconsistant with universe

[alaupin-ipsl]

In input4MIPs_CVs' main branch, names of 2 institution_ids are inconsistent with the DRS names of the corresponding terms in esgvoc:

• CNRM-Cerfacs => CNRM-CERFACS
• ImperialCollege => IMPERIALCOLLEGE

What do you think should be done @znicholls ? @ltroussellier suggested two solutions: either changing the names in input4MIPs (which would mean changing the datasets' names too), or overloading the DRS name on the project's side (which should be avoided to keep information centralized)

[znichollscr]

either changing the names in input4MIPs (which would mean changing the datasets' names too)

Not possible: datasets are already published and we're not going to republish

overloading the DRS name on the project's side (which should be avoided to keep information centralized)

I would do this: have two entries in the input4MIPs institution CV. One with the 'correct' spelling, one with the wrong spelling. Point them both at the correct universe entry. For the one with the wrong spelling, overload its DRS name and id (the id can be something like "imperialcollege-wrongspelling" ?).

Alternately, just don't include input4MIPs in esgvoc for now. Input4MIPs doesn't really work the way as all the experiment output projects, so we're going to crash into lots of headaches like this and it's not clear to me what the reward is right now (does it allow republishing on ESGF-NG?).

[ltroussellier]

does it allow republishing on ESGF-NG?

that is the intention

but more than that : the idea is to get the maximum information in esgvoc ecosystem in order to build tools that can link them. I dont really right now, but as a user if i download a dataset from esfg-ng i would like to know how it was produced, institution, model (the easy part), and also parent_* dataset (tricky part 😜) and also forcing dataset used. I know we are not there and we have other priority at the moment but to start little by little. having what we know about input4mip in esgvoc is the first step.

Right now, i dont see why it would NOT be possible : everything "works" in esgvoc for input4mip => CV queries, the DRS, The global_attribute, the STAC schema generation. The CV and specs are maybe not as much reviewed in esgvoc than other project but all tests are green.

I would do this: have two entries in the input4MIPs institution CV. One with the 'correct' spelling, one with the wrong spelling. Point them both at the correct universe entry. For the one with the wrong spelling, overload its DRS name and id (the id can be something like "imperialcollege-wrongspelling" ?).

Why not, @alaupin-ipsl if you still got some will left 😄