Consider requiring Java9 minimum (ideally 11)

[jimpil]

This would open the door to migrating away from clj-http, and instead rely on the native HttpClient (e.g. via hato which is a great little library).

Moreover, even if you don't like the sounds of that, the current http namespace could really use some love.

First of all, I am kind of confused by the use of many dynamic vars (are these intended to be re-bound often?), and a bunch of alter-var-root calls under connect!. I am probably missing something, but I 'm having trouble understanding the motivation behind this kind of pattern. From what I saw in various examples/tests etc, usages of connect! could easily be served by an atom instead. Anyway, it's just something that caught my eye...

Secondly, the way clj-http is used, it creates a brand new HttpClient for every single outbound request (some functions do two requests, so that would be 2 HttpClient objects). This is kind of wasteful, and if fact the clj-http docs do draw attention to that.

Anyway, I'm just probing here...let me know of your thoughts 👍

[michaelklishin]

I am all for requiring Java 11. Please submit a PR. We will likely bump the version to 6.0 after merging it, and should ship 5.5.0 before doing so.

[michaelklishin]

Addressed in #119.