I'll start: We're using async-graphql to power the API in Vector. We have two clients connecting to it:

I wrote a blog post about it here, although there are some updates required to that post to pick up the latest schema changes.

Things that have been particularly awesome:

The code-first approach for writing schema. In a previous project, we used a Go GraphQL library, which required writing everything in SDL and building resolvers around it. Being able to keep everything in Rust has been awesome.
Minimal boilerplate. Being able to drop in a few annotations around methods and parameters and get a fully working GraphQL API is perfect!
Relay connections. It was trivial to add pagination support to our queries, which has proven very useful when dealing with very large data sets. I love how this interface was designed - it was easy to come up with a cursor format that works well for our records, and the helpers async-graphql offers make it simple to roll out to everywhere we need pagination support.
Websockets + subscriptions. These are the backbone of exposing real-time metrics (and after enhancement(observability): Initial API for tap vectordotdev/vector#6363, events). It's so nice being able to return an impl Stream<Item = T> and it just work.
Above all, the incredible support we've received from @sunli829 personally when we ran into issues, or had feature requests. I don't think we've ever seen an issue go more than a few hours without being resolved. A million thanks 👍

I'm very excited by this project and where it'll go next. Thanks for making our lives (and my life in particular) easier!

Description

Metadata