Fetch record-definitions at runtime

[manuelbernhardt]

At the moment, record-definitions are built into a definitions.jar, then packaged together when the hub is built & deployed. So for each record-definition change the hub needs to be re-built and re-deployed.

Instead a better approach would be to have a kind of SchemaFetcher that would download the record definition and validation from the repository, for the starters perhaps only when the hub starts up (and later perhaps via an API call / UI button)

This would mean:

• implement a fetch class that downloads the enabled definitions and crosswalks (RecordDefinition.enabledDefinitions) from the definitions repository and stores them on disk
• in Global, call the downloading mechanism
• in RecordDefinition, replace getRecordDefinitionResources with something that lists the downloaded definitions from the filesystem

Some thoughts:

• we should start versioning record definitions / schemas. At least before starting source management this must be in place
• for versioned record definitions and such, there should be some sanity check via a hash to ensure that the same version wasn't changed

[glamcloud]

Great!!

And even better if we could see the current Field-markers (in reality a SOLR rec-def on top of icn, aff, tib etc) as a separate cross-walk-mapping and something being part of the source mapping. So fetch a given cross-walk using a another mapping as the source. Then all of it should be a bit more agnostic and future proof.