Something went wrong on our end. Please try again.
Created by: camdencheek
Previously, we had a method on searchResolver
(evaluateJob
) that translated any errors from a job into alerts. This converts that method into a job. This means that we basically have three, clean steps to searchResolver.results()
:
We're getting close to being able to decouple search code from resolvers entirely.
I don't expect this job to exist long-term. I think the ideal situation is that either 1) jobs emit alerts that apply to them, or 2) we generate alerts by expecting the execution results, and don't return alerts from jobs at all.
Should be semantics-preserving and covered by integration tests.