Hi there,

Glad that I found this. Was about to start the same as I ran into apollographql/persistgraphql#13 while trying to implement persisted queries.

I can't think of a more solid approach for solving this problem than a Babel plugin.

This is how I imagine it would work: We identify queries at build time, resolving fragments and generating a QueryDocument for it all. We can then use this QueryDocument for run-time operations (like replacing the query with an identifier in the networkInterface). But we can also use the exact same QueryDocument on the server in reverse. If the server is not in JS, you would have to print it out of course.

Even if you don't want the QueryDocument in your compiled JS (maybe due to the concerns that were raised in the issue on the graphql-tag repo) you could just print it out again while retaining the query identifier that was communicated to the server.

Does that sound about right?

I can offer help with working on this project. Maybe we should sync beforehand though as it's still in an early stage.

Really looking forward to this.

Regarding apollographql/graphql-tag#31

I started working on this sometime back and I'm not working on it anymore. I can add you as the owner of the package babel-plugin-graphql-tag on npm. I need your npm handle.

Thanks for working on this.