Initial mutations implementation

[dOrgJelli]

No description provided.

[Jannis]

@dOrgJelli Could we squash all of these commits into a single one, with a decent commit message?

[dOrgJelli]

@Jannis is this better?

[dOrgJelli]

Also side note: master is not currently protected.

[Jannis]

@dOrgJelli Now the commit message is too long 😉 How about "Initial implementation with Apollo support"?

[dOrgJelli]

Sure thing, will update shortly 😊 hopping on a flight to Denver ✈️ 🏔

[Jannis]

A few general notes:

• Let's drop the @client directive everywhere. Apollo warns if you use it wthout Apollo client-side resolvers.
• The tests checking whether useMutation and <Mutation /> set the observer object seem superfluous.
• The files in this repo are formatted inconsistently. Please run Prettier over all of them.
• Error messages sometimes end with a . and sometimes they don't. Let's remove all .s.
• Can we drop the _ on private / internal fields? Underscores are primarily used for indicating unused variables these days. Besides, they are not pretty. 😉

[dOrgJelli]

@Jannis

Let's drop the @client directive everywhere. Apollo warns if you use it wthout Apollo client-side resolvers.

From what we've found, when executing the query using ApolloClient, the @client directive is required. If it isn't the resolver will not be found. We thought about adding @client behind the scenes inside of the localResolverExecutor, but the document's directives are const/uneditable.

The tests checking whether useMutation and set the observer object seem superfluous.

I'd like to keep these as it's the only e2e verification we have for the state observer.

The files in this repo are formatted inconsistently. Please run Prettier over all of them.

Fixed :)

Can we drop the _ on private / internal fields? Underscores are primarily used for indicating unused variables these days. Besides, they are not pretty.

Fixed :)

[Jannis]

From what we've found, when executing the query using ApolloClient, the @client directive is required. If it isn't the resolver will not be found. We thought about adding @client behind the scenes inside of the localResolverExecutor, but the document's directives are const/uneditable.

That's because in the mutations package, you use the Apollo-specific withClientState. That's not going to work, as the mutations package is to be agnostic of any GraphQL framework. There is really no need for ApolloLink there.

[Jannis]

What really needs to happen for mutations to be agnostic of Apollo is:

• The schema is passed to createMutations along with the resolvers. We're not using the schema anywhere right now, which, as I am realizing now, is wrong.
• createMutations builds a GraphQLSchema object from the parsed schema and the resolvers.
• localResolver goes away and is replaced with a simple call to execute.

It is clear that this is too late for ETHDenver, so we're going to have to accept the @client directive for now. This is going to be an alpha release after all and we can still make breaking changes.

[Jannis]

I don't think this is needed for anything, so could be dropped.

[Jannis]

@dOrgJelli @namesty Something's up with Travis, I reckon the TypeScript config for CI is broken a little.

[Jannis]

Pushed a fix for Travis.

[Jannis]

I'm going to go ahead and merge it. CI will pass, apart from the book deployment for the docs, which are empty right now anyway.