Don't know much about golang for an API gateway it might be better than clojure, any thoughts? https://www.infoq.com/articles/api-gateway-clojure-golang

rewrites almost have to skew to the rewrite

@gklijs To be completely honest I would never expect Clojure to do better performance-wise on such tasks than Go.

I’m pretty sure most people don’t pick Clojure for performance, but for other reasons

Isn't an API gateway a highly event driven, reactive kind of application?

seems like node/js could be a decent fit there too

Obligatory https://www.youtube.com/watch?v=bzkRVzciAZg

lol

I agree with the comments re the grotesqueness of async code on node

"it was you" 😂

that was hilarious

Well, I'm a big fan of blocking ops, from a code layout perspective. But I doubt core-async can really be a full replacement for golang, in terms of bare metal performance for non-blocking ops, interleaving continuations and whatnot.

choice supportive bias, confirmation bias, expectation bias, illusion of validity, insensitivity to sample size, and post-purchase rationalization are all biases at work in that kind of comparison, which leaves out the big one, if you initiate a large rewrite that takes many person hours, you better have something to show for it

blog posts along the lines of "our company rewrote X in Y and Y is better" are interesting sometimes, but can never really convincingly argue their thesis

yeah that article was bad and they got super defensive on reddit about it

> The fact that the API gateway is now built in a typed language provides the ability to plug in diverse functionalities and introduce new technologies much more easily with Golang’s library support and community

that means literally nothing

Every rewrite benefits from the fact that you now understand the system better than you did when starting writing the previous version.

(speaking from experience as we're rolling out the 3rd gen of our platform which is essentially the 2nd complete rewrite from the ground up!)

Yeah, actually, I didn't understand the reference in the conclusion of 10X improvement, given the performance comparison chart

The avg latency improved by 50 or 60%, with the complete rewrite