Apollo Router vs. Cosmo query plan execution?

[rhino88]

Is the expectation that the Cosmo router will have exact parity with the Apollo router?

For example, we have discovered an overfetching issue in the Apollo router which significantly increases the latency of some of our federated queries in production.

We have confirmed that this issue also exists in cosmo/wundergraph. Would this be something that Apollo needs to fix before Cosmo would do the same?

[devsergiy]

Hi @rhino88

We are not exposing yet human readable query plans

But there is a way to print internal plans by providing this environment variable

 ENGINE_DEBUG_PRINT_QUERY_PLANS=true

Could you run the router with your configuration and paste query plans here?

[rhino88]

Sure thing, I also realized that I created the example using the Apollo Federation quickstart which does not use cosmo. Will publish an example asap.