Clojurians Log v2

What problem did you have with the port setting?

Btw, do you happen to have piggieback on your classpath for some reason in those pure Clojure projects? After looking a bit at the code this seems to be the reason you’re getting such warnings after the recent nREPL changes and that’s actually the expected behavior.

That's explained in the issue that I linked and there we are also discussing with @alexyakushev what's the best course of action for this.

@bozhidar Yes, you are correct, piggieback is being pulled on my classpath even in a pure Clojure project, which is unexpected and I am investigating why.

@bozhidar Doesn't that raise another question, though: if piggieback is on the classpath, what is this warning warning against?

It might be the problem with clojure-maven-plugin. The warning did repeat for me in your repro example. Could be the case with misconfiguration. For example, I can't make the port setting to work either.

@alexyakushev @bozhidar Of course, here it is: https://github.com/danielsz/piggieback-warnings Please do not hesitate to get back to me if you have trouble with anything.

The warning is against having piggieback on classpath but not explicitly including its middleware in the nrepl server being started. It used to be included automatically by cider-nrepl, but it is no longer the case.

Therefore, I think the current behavior is reasonable. Removing piggieback from classpath should get rid of the warnings, which is a viable solution.

I agree with the principle of piggieback not in classpath -> no warnings vs in classpath but no middleware added in middleware vector -> warnings. Unfortunately, this does not describe the behavior with https://mvnrepository.com/artifact/com.theoryinpractise/clojure-maven-plugin where the warnings appear consistently despite piggieback being in the classpath and its middleware being present in the middleware vector.

That shouldn't happen. I would appreciate a small reproducer

I'll take a look, although at a glance it seems to me that's the expected behavior right now. (at least as far as nREPL is concerned) The warnings are actually caused by the way cider-nrepl use piggieback internally, and probably the real fix will happen there.

See also https://github.com/clojure-emacs/cider-nrepl/pull/971

I have not. Maybe running ssh over the serial port with port forwarding might be easier? Would that be possible?

Porting SSH and an IP stack to these microcontrollers seems like a lot.

I guess the IP stack isn't necessary, since you said over serial...

In my case, this is for microcontrollers, such as the Pi Pico or Arduino.

The protocol already supports this, the session id is part of the message. The degree to which clients support this and middleware respects it, I dunno