Update graphql to 0.5.0, closes #130

[wenzowski]

next release of graphql has breaking changes
see graphql/graphql-js@d7cc6f9

not ready for merge

[wenzowski]

app boots with these changes and appears to work on a quick perusal but the test suite fails on a rethinkdb call I don't have time to track down right now

[mattkrick]

nice! let's give it a couple days to hit npm, npm still doesn't treat github repos right (3 years and counting! npm/npm#3081), also it doesn't allow for local caching, which kills if you're installing from a country with less-than-great internet.

[wenzowski]

yeah npm + git = 😢
...thus the tarball 😉

graphql actually isn't usable directly from git anyway since @leebyron changes the build dir in his prepublish hook

[mattkrick]

wowwww, if anyone wanted to compete with npm, now would definitely be the time for it. npm3 10x slowdown + left-pad + a book of unresolved issues...

[wenzowski]

definitely time for gx

[mattkrick]

well, at least the name is better than ipfs 😄. looks like it's getting a lil traction in JS land, too: https://github.com/ipmjs/gx-node

this could get interesting...

[wenzowski]

rebased and bumped graphql dep to latest

[mattkrick]

Looks good! Just 1 question... in the spec, the param for authToken is called context. Should context be an object that holds the authToken, or should it be the authToken itself, as it currently is? Basically, this: graphql/graphql-js#326 (comment)

I like how clean it is now, but the context will need to hold other things (for example, in the subscriptions the socket probably belongs in context instead of refs.rootValue since we're literally using the context of the socket & rootValue is more like the surrounding calculations). Thoughts?

[wenzowski]

Very good point. I'll look into that, and likely move the socket in too.