Watch the replays: GraphQL Working Group Meetings on YouTube
- Lee: looks like schema coordinates is a separate grammar (so that we don’t have to have questions about comments, whitespace, etc…)
- Benjie: different syntax, but same tokens.
- Lee: we already have dot punctuator, which can be handled separately for schema coordinates.
- Mark: function on AST nodes? Used for error messages.
- Benjie: Do we need a separate appendix B? What does that mean for “Name”?
- Lee: “Name” is already a lexical token.
- Benjie: looks like we need a separate lexer implementation?
- Lee: yes.
parseName()can probably be reused. No String values. No complicated lex tokens. - Lee: it’s also good for tree shaking/robustness
- Mark: feeling good about the graphql-js changes
- Lee: I can do a PR on your PR about the graphql-spec
- Mark: is it still OK for the spec cut?
- Lee: it may still be good?
- Benjie: after the error has propagated, no other error can propagate
- Lee: parallel execution: 2 errors are encountered
- Lee: it’s non-deterministic
- Benjie: that is fine
- Lee: we could add “propagationDepth”
- Lee: iOS engineers wanted to avoid drawing partial object
- Lee: for fields, we group per arguments/directives
- What if we add a similar rule for fragments?
- Benjie: that doesn’t work with components oriented design
- Lee: we break that already for fields. Fragments are not modulazired.