Righ, jack-in works fine, because the starting point is the current project so it loads the deps.edn file properly. If I start a repl with this clojure alias

:cider-nrepl {:extra-deps {cider/cider-nrepl {:mvn/version "0.21.1"}}
                :main-opts ["-m" "nrepl.cmdline"
                            "--middleware" "[cider.nrepl/cider-middleware]"]}

So do you manage a shared .lein/ profiles.clj file in shared version control system, or do you add tool dependencies in your projects?

Can you elaborate why the REPL port needs to be the same? And how do you keep it the same, I thought it was random

Data point: I use fixed ports for convenience because they are tunneled

I add tool deps in my project.

:nrepl
           {:extra-deps  {nrepl             {:mvn/version "0.6.0"}
                          cider/piggieback  {:mvn/version "0.4.0"}
                          cider/cider-nrepl {:mvn/version "0.21.1"}}
            :extra-paths ["dev"]
            :main-opts   ["--main" "nrepl.cmdline"
                          "--middleware" "[cider.nrepl/cider-middleware]"
                          "--port 9000"]}

And the ports are fixed, because the Cursive Run configuration is also version controlled. This way, a new person on the team can checkout the code; make repl; and connect via Cider or Cursive, consistently.

So what you suggested above, you start a repl separately, then do cider-jack-in or connect?

Oh I see now. You have the alias for nrepl in your project and just kick it off there from a script, so you get the deps for your project but it's not tied to emacs

I think my problem is I'm kicking it off from my user deps.edn so it has no dependency information, this makes more sense

I start nrepl with the above in the project deps.edn yes, and clojure -A:nrepl - then cider-connect-clj to it.

I like the stability and the explicit version numbers

right, I got you, that make sense. My perfect use case would be to have one "Central" repl that I can just connect any application into and dynamically add deps to the classpath upon connect, so then I will never have to worry about opening a new connection or if I'm working on new projects,I can start it up from my profile.clj on startup

but seems java gets in the way of that

I want my repl’s to be project specific

To complete the picture, I also have this:

(cider-register-cljs-repl-type 'user-custom "(start-cljs-repl)")
  (setq cider-default-cljs-repl 'user-custom)

hmm, I'd have to look into that, I am on and off cljs and use shadow, but mostly on clj so not sure what that is doing

Swapped from shadow to figwheel last week, and my Emacs remained unchanged - it just connects to whatever that function does.

Now:

(defn start-cljs-repl []
  (let [build-id "common"]
    (do
      (try
        (figwheel.main.api/start build-id)
        (catch RuntimeException e
          (figwheel.main.api/cljs-repl build-id))))))

Then:

(defn start-cljs-repl []
  (shadow/watch :local)
  (println "Open  to connect and start an evaluation environment.")
  (shadow/repl :local))

Same workflow, from the REPL perspective

is there an easy way from emacs to run the aliased command in your project on the host rather than using cider jack in?

I'm thinking I can just make a function in elisp for it