Caching options on connected APIs

[chris-schra]

I just discovered wundergraph and it looks really awesome, plus the website describes the features so well. Thanks for that.

One thing I haven't found yet is an explanation how caching works between wundergraph and the remote APIs.
From the docs:

const operations: ConfigureOperations = {
    ...
    queries: config => {
        return {
            ...config,
            kind: "query",
            caching: {
                enable: false,
                public: true,
                maxAge: 10,
                staleWhileRevalidate: 5,
            },
            liveQuery: {
                enable: false,
                pollingIntervalSeconds: 5,
            }
        }
    },
    ...
    custom: {
        Countries: enableCaching,
        TopProducts: config => ({
            ...config,
            caching: {
                ...config.caching,
                enable: true
            },
        }),
        Users: config => ({
            ...config,
            caching: {
                ...config.caching,
                enable: true,
                maxAge: 120,
            }
        }),
    }
}

I guess these configs set the caching behavior between clients and wundergraph. But what about the caching between wundergraph and APIs? Especially when using live queries...

[jensneuse]

@chris-schra You're right in that this configuration defines the HTTP caching behaviour between WunderGraph server and clients.

Currently, there's no caching happening between WunderGraph server and origin, although it would be simple to add.
E.g. the transport we're using to talk to the origins could easily support caching too.
It's just not something that anybody asked for, yet.

Regarding live queries and origin caching, how could this help in your use case? Can you describe a bit more what you're trying to do?

[chris-schra]

@jensneuse regarding live queries: we „offload“ etag caching to Cloudflare so basically API calls routed via CF would consume no resources and we could easily use a very low polling interval.
Plus, but that‘s a far future, I‘m thinking of dependent caching - like only query this API when other APIs response is stale