would it make sense to have some kind of --reload option for clj-run?

I'd suggest using clj-repl and invoking the function via the REPL while developing it

you can do (require 'the.ns :reload) then

cursive and others let you Load file in REPL directly as well which also works

I feel like the REPL would be much more productive than --reload

but I can add it if you feel like you need it

hmm

while i am developing, i’d be using a repl (probably evaluating the stuff from my editor)

but sometimes i might have a process running in the background from coding, and i want to build a release from the command line using the latest state of the code

any particular reason you are not calling that from the REPL?

i have a release command in package.json so that other team members only need to remember one thing (eg. npm run release) and this is what our ci / build systems call

ah. makes sense.

I feel like you'd just add --reload permanently then though and just let it always reload?

yes

the intent would be that anyone else touching the system wouldn’t have to worry about npm run release not using the latest version of that code

I don't think clojure records when something was loaded. otherwise I could probably do something automatic

by way of example, say i have a dev-server running in the background. i run shadow-cljs watch browser and do some front-end coding, with live reload. Then I close the watch process and want to run a release

normally i’d just run this from the command line, npm run release, but since the dev-server is in the background i end up restarting it whenever i do releases. possibly because i just haven’t developed the habit of using a repl for that bit

hmm yeah restarting the server shouldn't be a necessary thing

I'll see if I can figure out an automatic way to do this. if not I'll add the flag.

:thumbsup:

is there any difference at all between shadow-cljs watch browser vs. shadow-cljs clj-run my-build-ns.core/watch, wherein I do some cleanup and then run (shadow.cljs.devtools.api/watch <build-id>) ?

(where ‘cleanup’ means ’revert some asset-path transformations performed during release` and other such build-pipeline-related stuff)

tiny bit of difference maybe

watch stops the build when connected to the server an you CTRL+C

the clj-run probably doesn't

I'm still sorting out the hook API that would let you hook into :watch-start :release etc events

then you could probably use that instead of clj-run

hmm ok

so in the meantime, shadow-cljs clj-run my.build.ns/before-watch; shadow-cljs watch browser;

which, if a server is already running, is super fast, otherwise it starts a server process twice

shadow-cljs start; shadow-cljs clj-run my.build.ns/before-watch; shadow-cljs watch browser;

start starts a background server if none is running

ah great.

do you recommend putting shadow-cljs into :npm-deps?

no

at most it is a dev dependency which :npm-deps doesn't handle