This will be just a simple anouncement: I’m publishing my (still very crude!) plug-in for Clojure’s Socket REPL on Atom.
Chlorine was an experiment that grew more than I could imagine: I wanted a better way to write ClojureScript on Atom, and that’s the reason I made Clojure-Plus (also, I wanted to seamlessly integrate CIDER and other refactoring/helper libraries on my workflow). But there were lots of bugs, problems, with my approach: I was using CoffeeScript (the default language on Atom packages, at the time) to write my plug-in. This meant lots of conversions from Clojure to JS, and every single bug I needed to reload atom to re-activate my plug-in.
Then, I tried to work with Figwheel. Atom packages are a mix of Node.JS with Browser target: figwheel was simply a lost cause (lots of problems when reloading when it didn’t knew what to reload, inconsistent state, flaky compilation failures that sometimes did work, sometimes not, without any logic). When I migrated to Shadow-CLJS, things worked a lot better.
Then, Microsoft bought Github. Atom always had a bunch of things that I did not like (quasi-semantic versioning, lots of feature requests that were simply ignored, multiple bugs that were not being addressed, performance issues) and with Microsoft buying Github (and already having an electron-based editor – VSCode), I’ll not wait to see Atom becoming obsolete. Most of the people that I work with don’t use Atom anymore, and it’s kinda difficult to see people working with Clojure with Atom too, so I started to think on a better way to port my work to other editors.
Then, I ended up too far on the rabbit hole: I lost track multiple times of what I was doing on REPL Tooling. So, for now, I’m developing both packages in parallel (if you look at Chlorine repository, you’ll see a submodule registered) so that I could code a tool, integrate immediately on Atom, then see the results in real time. This became a much better workflow (at the expense of unit tests for now). And then, one day, I was working on a ClojureScript project on my current job, and after multiple auto-completes, evaluates, and other things, I realized that my plug-in was good enough to be usable (and in a way that I would prefer to use it instead of other solutions).
So, I’m publishing it. Expect bugs (lots of it!), but it’s a start. I’ll try to release every new feature as soon as possible (possibly in an automated way – who knows?), and there are still multiple things to do (go-to definition, for instance, is completely absent) but I’m sure that there are lots of interesting things to happen!