I've tried an experiment with Planck where it effectively understands deps.edn
.
What it is really doing is, if classpath information is not explicitly provided via -c
/ -D
, etc., it then just shells out to clojure -Spath
to get the classpath.
This seems to work out nicely... I'm also thinking it might be nice to allow pass-through of options to clojure
, like -Ralias
, -Sforce
, etc. Interestingly, neither Lumo nor Planck have used -J
, -R
, -C
, -S
, so there is no conflict, but there might be a better way to pass through than to hope for no collisions.
Mentioning here in case any Lumo maintainers are pondering something similar.
If curious, the Planck changes are in this branch: https://github.com/mfikes/planck/tree/cli-tools
I like that we keep dep resolution separate myself, because I can choose better strategies (custom boot tasks with different profiles for instance) for it. My 2c, in any case it is definitely worth considering the option as it would streamline that process for sure.
Yeah, right now you can compose lumo and clojure via
lumo -c`clojure -Spath`
so all I really did in the Planck experiment was to compose them automatically. Pros and cons with that idea.I guess the decision will be mainly driven by editor adoption...having deps.edn
there will really make things easier for editors...anyways thanks for sharing!
(as always 😸)