@Jon so my “solution” to this is just having a server running. just do a comparison. a)
time yarn your-scriptsb) runyarn shadow-cljs server, in another terminaltime yarn your scripts. The difference should be substantial when the server is running since you only pay for startup once. (edited)server management itself is a bit annoying at the moment but I’m working on improving that
I tried. The time was cut from 10s+ to 2s in one of my project with caches for compilation. In many occasions I would release my projects with the server running.
so there might be a
shadow-cljs start|stop|restartat some point which starts a background server so you could modify your scripts to first callstart, then do all the rest, then callstop. (edited)
I would start a watch server and keep it running actually. I think I would rely on the logs of the server to read its status. If it runs in backend, I can’t see the logs.
besides that I can make
shadow-cljs clj-runstart as fast asclj, it is slow since it wants to provide easy access to shadow-cljs itself. as soon as you are going to use anything you are going to pay for that startup time again. UNLESS you are using the server instance …
That would be nice. My most frequent use case of Clojure REPL is to try macros, or I would just use Lumo. Besides, now it’s 4 keystrokes away to start a Clojure REPL.
installing shadow-cljs takes about 20sec for me with the new AOT enabled, less than 5sec wit
Bad network. Would you mind adding a processbar?