Today I saw this: https://github.com/stickfigure/blog/wiki/How-to-(and-how-not-to)-design-REST-APIs
And it reminded me why GraphQL is so much better.
It has also reminded me why HTTP is so confusing and awful as a protocol, especially as a protocol for structured data APIs, with all its status codes and headers and bodies and querystrings and content-types -- but let's not talk about that for now.
People complain about GraphQL being great for frontend developers and bad for backend developers, but I don't know who are these people that apparently love reading guides like the one above of how to properly construct ad-hoc path routers, decide how to properly build the JSON, what to include and in which circumstance, what status codes and headers to use, all without having any idea of what the frontend or the API consumer will want to do with their data.
It is a much less stressful environment that one in which we can just actually perform the task and fit the data in a preexistent schema with types and a structure that we don't have to decide again and again while anticipating with very incomplete knowledge the usage of an extraneous person -- i.e., an environment with GraphQL, or something like GraphQL.
By the way, I know there are some people that say that these HTTP JSON APIs are not the real REST, but that is irrelevant for now.