preview @defer support

[BrynCooke]

Current plan

Implement defer only for now, in the router, with some query planner modification.

@defer RFC

Related issues:

• Defer: split queries into primary and deferred queries for format_response #1178
• Defer: return a stream of responses from RouterService and ExecutionService #1179
• Defer: implement query plan execution for deferred nodes #1180
• Defer: response encoding #1181
• defer: add a map_deferred_response in rhai #1219
• rhai: add a map_deferred_response on ExecutionResponse and RouterResponse #1409
• specify the path instead of creating a deferred response from the root #1469
• indicate the supported defer specification version in Accept and Content-Type #1648
• defer: add a map_deferred_response in rhai #1219
• Update to Federation 2.1.1 #1666
• use a HTTP 406 (Not Acceptable) response when the Accept header does not match the defer query #1659
• rhai: add a map_deferred_response on ExecutionResponse and RouterResponse #1409
• update the embedded sandbox #1426
• specify the path instead of creating a deferred response from the root #1469
• the @defer directive is not returned by introspection queries #1544
• update to federation v2.1.0 #1546
• Align @defer implementation with latest proposed spec updates #1548
• deferred responses should be compressed #1572
• the content type should be application/json if there's only one answer #1575
• supergraph-demo smoke test 7 fails on router 0.16 #1584
• Defer: send the multipart boundary at the end of each part instead of at the beginning? #1588
• add a spec compliant parser for Accept HTTP header values #1618
• router-bridge: Depend on directly-depended-on deps rather transitive deps federation-rs#160
• router-bridge: Update to now-published federation 2.1.0 packages federation-rs#161
• Response to Deferred Query with Accept: multipart/mixed is application/json #1656

To be defined:

• response stream format: right now the stream is apparent in all signatures, maybe we could have instead response types that holds the headers then a (generic) stream of graphql responses
• adapting telemetry: telemetry now registers one operation per response (deferred or not), that might not be what we want in the end. But also, telemetry does not really have a concept of deferred responses

[abernix]

I've changed this issue to just be about @defer. We can still think about @stream support in that design, but the outcome might just be @defer. 😜

[abernix]

This is done via a lot of tickets.