how are you starting it? all the regular commands already detect and connect on their own?

I am starting it from a Phoenix watcher. So when the Phoenix server starts, it automatically starts. The issue is whenever I kill the Phoenix server, the shadow-cljs server does not die, hence, when the Phoenix server starts again and tries to spin up shadow-cljs, it returns an exception about a server that's up.

why doesn't the phoenix server kill the shadow-cljs instance when shutting down?

again though: what command do you use to start shadow-cljs?

if its through the regular shadow-cljs command it should already connect just fine. unless phoenix killed its runtime files in the .shadow-cljs dir

maybe you should try a separate tool to run the phoenix server and shadow-cljs in parallel. if phoenix isn't capable of killing the processes it started that doesn't seem great

I use shadow-cljs watch app to start the server.

that'll figure out if a server is running and connect to it

unless its runtime files in .shadow-cljs are deleted by something else

What is its runtime file in .shadow-cljs, let me confirm that.

there are files in that directory that the command will check to see if a server is running

also caches and stuff. so the entire dir should be left alone and never be touched by anything but shadow-cljs

no clue what phoenix does, just a guess

Alright, thanks.

I'll just try out some things here and see how it goes.

the correct solution should be to just have phoenix kill shadow-cljs when it gets shut down

leaving a lingering process is not a clean solution in either case