Proposal: `neil add shadow-cljs` about neil

[borkdude]

@rads I think a solution to the above could be a neil new shadow-browser-app template or so.

[borkdude]

I'm using shadow-cljs in a couple of projects myself and I find neil add deps thheller/shadow-cljs sufficient for now. Taking on all the options that shadow-cljs provides is imo too broad of a scope for this project. Also it's not always the case that you want to bundle the assets that shadow produces into your uberjar, if you create an uberjar at all.

[borkdude]

I feel it's more the job of shadow itself to create such .edn files based on command line options.

[borkdude]

To add, I'm using shadow-cljs mostly for testing or producing libraries and publish them to NPM. So that doesn't align at all with the above proposal for example.

[rads]

@borkdude: I totally get where you're coming from. I should clarify that I want to solve a specific problem here: I believe ClojureScript is too hard to set up and use to build web apps in 2022. Disclaimer: This is just my opinion and I'm incredibly grateful for all the work that the community has provided to us for free.

Nonetheless, there's glue that's often missing and my dissatisfaction with the developer experience motivates me to solve the problem however I can, whether that's through neil or shadow-cljs or something else.

For example, understanding what goes in all those files above just to get started was an annoying process for me even though I've been writing ClojureScript professionally for 5+ years now. The steps need to be performed manually if you want to add these things to an existing project instead of starting from a template. I know people who could really enjoy ClojureScript but may not be able to get past all these little steps on their own.

Anyways, rant over. This project is awesome and I want it to stay that way. With all this in mind, I see a few possible approaches for me to move forward:

1. Document usage of neil add deps thheller/shadow-cljs and the above files
   • Not enough of a solution from my point of view since it requires a lot of manual understanding to build a simple web app, but it could be a good starting point for the following options
2. Add improvements to shadow-cljs itself so the wrapping isn't needed
   • Probably the best in the long-term but will likely require a lot more time and effort to get changes inside shadow-cljs itself than it would to get into a wrapper tool
3. Create a separate tool from neil to implement what I described above
   • This seems like a good way to experiment for now, though it means there would be another tool for people to install and learn

Since this isn't really relevant to neil anymore, I can go ahead and close the issue.

[borkdude]

And perhaps host it with shadow-cljs so it can be used as neil new thheller/shadow-cljs ... or so?

[borkdude]

Or we could still have a neil add shadow-cljs --browser thing perhaps. I just found the above a bit too much. E.g. adding things to build.clj could be a bit too interfering.