search

WCRP-CMIP / CMIP6Plus_CVs

Controlled Vocabularies (CVs) for use in CMIP6Plus
Creative Commons Attribution 4.0 International
3 stars 4 forks source link

Searchable Facets for ESGF Client Applications? #51

Open sashakames opened 6 months ago

sashakames commented 6 months ago

The CVs could be used as the single source of truth for applications but they need some additional metadata. One example is whether a Global Attribute or "extended" attribute exists as a search facet. If this information is not provided, we would need to maintain it elsewhere, eg. the esgf-config repository, then there would be multiple sources which is not idea.

sashakames commented 6 months ago

Eg, We could have either a distinct .json file or this would be included an existing json file. Eg the Global attributes could look like this:

[ { 'name' : '<att-name>' , 'search-facet' : <True|False> } , {} , ... ]
durack1 commented 6 months ago

@sashakames thanks for chiming in with this comment. It might be useful to migrate this to the CMIP6Plus_CVs repo to start, as this project is under active development, whereas CMIP6 contributions are starting to slow down markedly. Are you happy for me to migrate this?

sashakames commented 6 months ago

One concern is that existing projects will need the additional metadata. So we could migrate the issue, but something will be needed for CMIP6 at some point, and soon as we (ESGF2 folks specifically) are developing client applications that are figuring out what search facets to use.

sashakames commented 6 months ago

Also, are we (PCMDI/WIP people) planning to create "legacy" CVs for old projects, eg CMIP3/5? If not, could ESGF2 people do that just to keep our apps happy?

durack1 commented 6 months ago

@sashakames that would be the ideal, but will require some archeology to get details correct for older model configs (especially CMIP3). At this moment, the intention is to prioritize the "new" projects generating data (e.g. CMIP6Plus), but once that is live, and functioning as intended then backfilling CMIP5_CVs and CMIP3_CVs could be done.

I'll migrate this to the CMIP6Plus_CVs repo so we're keeping the discussion live

sashakames commented 6 months ago

Agreed that CMIP3 is highly-probably low priority, but CMIP5 is presumably still more active right now (from a search and use perspective) than is CMIP6Plus, where data will take some time to percolate up. So for the existing clients they will need sufficient data to test. So at least I'd hope we might be able to prototype something on top of CMIP6 soon, then ESGF2 could take a stab at doing the CMIP5 backfill so the application would be usable for that data as well