Ben Sless

@greglook I think so, but if you don't mind, a couple of points related to this design: - Should the client be passed as an argument? What if users want...

I wasn't exactly imagining a unified callbacks interfaces, but that isn't mutually exclusive. I thought of a mechanism where you could create a client with a default callback for `on-error`,...

My idea in particular was to spec `authenticate!` itself https://github.com/amperity/vault-clj/blob/master/src/vault/core.clj#L12

Having given it some thought, I may have an idea, it also might coincide with needs for open telemetry or other metrics frameworks; Add an interceptor interface with three methods:...

@ptaoussanis that issue is a good step in that direction but we'll still need some way of injecting interceptors around request handling

Rough starting point, but courtesy of @borkdude ```clojure ;; .clj-kondo/meander/epsilon.clj: (ns meander.epsilon (:require [clojure.string :as str] [clojure.walk :as walk])) (defn collect-vars [expr] (let [vars (volatile! [])] (walk/postwalk (fn [x] (when...

That's certainly better, however, what about clojure CLI default aliases? I usually have it set to `dev:test`, would letting users choose their aliases collide with that? What about global aliases...

That sounds like the most ergonomic solution: users can specify their defaults and CIDER will prompt them to choose. Sounds great

I see your point - I should perhaps qualify the code path by which I observed it, which was the request diff middleware from malli -> deep-diff2 -> fipp. That...

Ideally Clojure would be faster, although if we take a more philosophical approach, one never needs call `satisfies?` as any use of protocols outside of markers (first time I'm seeing...