Apply a schema to the holding model

[jcoyne]

This helps us enforce the shape of the data and highlights the fields we think are important.

[cbeer]

I'm pretty -0 to this. It feels like a pretty complicated solution where either not using fetch or providing a default argument would seem to do just as well.

I also think there's value in keeping our code pretty close to the FOLIO data structure for talking to our colleagues in libsys + metadata about where/how we're mapping data. The abstraction seems like forces a deeper understading of the code, doesn't it?

[jcoyne]

I also think there's value in keeping our code pretty close to the FOLIO data structure for talking to our colleagues in libsys + metadata about where/how we're mapping data.

Indeed. I did this by looking at the RAMLs that folio publishes. These data structures are derived from that definition.

quicktype -s schema ramls/holdingsrecord.json --namespace FolioTypes -o types.rb

[cbeer]

Interesting. I feel like I need to know more about what it's doing.. the camel case -> snake case is fine, but suppressFromDiscovery turning into discovery_suppress, location/effectiveLocation getting collapsed, etc surprise me.

[jcoyne]

suppressFromDiscovery was not in the raml model. That is something we extract from the jsonb in the query, so I had to add that myself. I collapsed effectiveLocation as that is the only location that we're currently using.