Renée

Thanks for the notification! If the upgrade is easy enough I'll just add it to the range of supported versions of this package :) At first glance it looks like...

How so? Looking at it again, I can work around it by telling the test suite about the version in a different way (by having the github action provide it...

2.1.0 is out with v5 support 😄

We have a hardcoded introspection recursion limit now that should prevent problems. I'll close this but customers that desire narrowing introspection down to a single hardcoded query should talk to...

Maybe instead of `get_line_column_start` and `get_line_column_end`, we could have a method that returns a `(GraphQLLocation, GraphQLLocation)` tuple? This would address the double-lookup and I think it would be extremely rare...

Would be nice to add a test for a subgraph that responds with non UTF8 headers.

Actually it seems like non utf8 headers are fine. it's probably about the messages themselves? I don't see a _nice_ way to test that and I think we should revisit...

Repro in https://github.com/apollographql/apollo-rs/pull/917

The reason appears to be that the `executable::Field`'s subselection gets emptied during ast -> executable conversion, and validation does not know about this

Yeah I've also noticed this a bit. I think I agree it should be removed. Maybe it can do something else to more specifically address the common "ARTIST 'Song Name'...