tuxmain Right now, the Pusher extension is tightly coupled to the specific implementation of the pusher service. In the long term, I'd like to decouple the two. This would mean refactoring Pusher to a "realtime" extension, where pusher would be just one driver. Other extensions could then extend 'realtime' to add new drivers. This way, the user-facing interface would be consistent (no more bugs like with websockets on Discuss), and extensions seeking to replace pusher wouldn't need to reimplace the wheel.
I don't think I've put together an issue for this yet, I'll do so now.
tuxmain Do you think someone with no experience with development of Flarum extensions, but with some experience in JS/PHP, can fork the flarum/pusher extension to make it compatible with a homemade websocket server? (which I would write in Rust)
As to this, I believe that most of the trouble people have with developing for Flarum comes from understanding modern PHP practices, plus several overly complex areas of Flarum (stay away from visibility scoping unless you're willing to dive deep into source code and inspect generated SQL queries). I don't think the Pusher stuff is too complex. If you have dev experience, picking up basic Flarum development should be relatively easy.
