How and why GraphQL will influence the Sourcehut alpha

I like my front end team to not write queries.

I have seen too many front end developers write queries equivalent to "select * from users, t1, t2, ... tn, left outer join on users.id = t1.user_id... etc etc etc".

I think that is just bullshit.

I prefer to tell the frontend people - these are the endpoints that backend team provides, go live with it. If you really have issues, talk to them, and they might just add a new endpoint for you.

If you let frontend developers dictate backend queries, all you will get is a big master cluster fuck. I am talking average joe developers that we usually find to work on startups.

I think Frontend has the right to take over certain domains of the backend. A very clear example of this is templating. Templating used to be a backend task, but it became very clear it was the frontend developers that needed power over this domain. Going over to a backend dev to make adjustments to the templates is an inefficiency that had to be solved, period.

I agree with you that writing optimal queries or data modeling should not be shifted over to the frontend. With that said, there are basic aspects of this equation that can be factored out. There should be a layer where frontend can at the very least pick or mix the data they need. That doesn’t necessarily mean Graphql, but could also mean a simple middle stack layer where one can do this basic thing without being able to shoot themselves in the foot.

There’s a responsible way to do this, and most likely will be an ongoing discussion.