I'm just wondering what's the reason for this seemingly awkward configuration (from Getting Started w/ Apollo Server),
const server = new ApolloServer({
// These will be defined for both new or existing servers
typeDefs,
resolvers,
});
server.applyMiddleware({ app }); // app is from an existing express app
Why is that I'm calling .applyMiddleware() and feeding it my app rather than using app.use(), it even seems from the docs that Apollo is only answering requests on /graphql wouldn't it be better to follow the Express API of,
let apollo = require('apollo-server').ApolloMiddleware
app.use( '/graphql', apollo({ typeDefs, resolvers }) );
It seems like Apollo is inverting the normal middleware flow of Express? What is the advantage of doing it the Apollo way?
I don't work with those guys, but based off the code that is there, I assume it has to do with the fact that it adds body-parser, upload capability, websockets (if you opt into this one) to the app, as well as a .well-known configuration for engine's health checks. The well-known would have to be at root according to that spec. Additionally, if you're using the subscriptions and websockets, they listen for httpServer.once('listening'), so they have to have access to your app. If they simply handed you one that you could mount, you wouldn't app.listen on it, since you'd app.listen on your own instead.
Just from the internals, this is what I would expect, at least.
来源:https://stackoverflow.com/questions/51716013/why-isnt-apollo-server-a-middleware-for-express-rather-than-being-a-server-tha