I am curious if you have any guidance on the developer experience when it comes to working with federated graphs. It seems like Apollo Studio will be playing a critical role in local development for the live reloading of the gateway. Otherwise, it seems a developer would need to stop the gateway, run the rover command against the subgraph, then restart the local gateway. The demos and walkthrough seem to utilize rover constantly throughout the developer experience. Previously, it felt like for development it leverage a lot more just-in-time introspection to build the supergraph.
Hello! It is true that part 1 of the Federation Quickstart focuses on using the Rover CLI to compose a supergraph schema, similar to the process that Apollo Studio uses when you enable managed federation in part 2.
The just-in-time introspection method for composition that you describe is still available as well, via the
serviceList option you provide the gateway (see the docs). However, this method does still require (and always has required) restarting the gateway whenever you modify a subgraph schema or add or remove subgraphs.