caching the result of an expensive computation/rpc-call in a GraphQLResolver

[kurtseebauer]

Hi all,

I have a GraphQLResolver like this (in Kotlin):

class MyEntityResolver : GraphQLResolver<MyEntity> {
 fun propertyA(e: MyEntity) = e.propertyA
  ...

and then some of the graphql properties are calculated in an external service call, like

  lateinit var service: ExternalComputationService

  fun propertyX(e: MyEntity) = service.compute(e).propertyX
  fun propertyY(e: MyEntity) = service.compute(e).propertyY
  fun propertyZ(e: MyEntity) = service.compute(e).propertyZ

Now my question is, how do I best store the service.compute(e) as a cached result? Do I put it in the graphql context? Do I have to deal with double checked locking and stuff like that myself or is there a better way? I want to avoid computing multiple times, and I want to avoid putting transient fields (maybe a by lazy property) in the jpa entity.

Thank you for any hints!