Supporting different frontends

[vynulldev]

Hey all!

Something I want to put out there as an idea, because I don't think it's obvious from the outside.

Vynull isn't really a web app. It's a headless daemon with a full HTTP/JSON API and a CLI, and the browser UI that ships with it is just one client talking to that API. The now-playing overlay and the MPRIS integration are already two more clients consuming the same state. So the interesting thing is: the API is the actual product, and the UI is replaceable.

Which means anyone could build a completely separate frontend against it, without touching the core at all.

A native desktop frontend

The one I'd most like to see, and the clearest expression of the idea, is a native desktop app that you compile and run, no browser involved. A real GTK app (or Qt, or a Rust GUI like egui / iced / relm, whichever), built against the API, shipped as a binary.

The part that makes this genuinely different from a bundled desktop app: because it talks to the daemon over the network, the frontend and the daemon don't have to be on the same machine. Run Vynull headless on a Pi or a little box in the booth, and drive it from a native app on your laptop across the room. The engine, the NFS serving, and the Pro DJ Link stay on the always-on box; the GUI is just a window into it that you can open from anywhere on the network, or close entirely without the music stopping.

So you'd get the native desktop feel people like, without the core having to become a desktop app. That separation is the whole reason the headless design exists.

Other directions

• A phone remote. Browse the library and load tracks from your phone on the same network. The search / load / players flow is already there.
• A booth or venue display. A big, read-only now-playing screen driven by the live status and beat feed.
• A different web UI entirely. The shipped one reflects my taste; someone who wants a cleaner or more opinionated design could build their own against the same endpoints.
• Hardware / Stream Deck integration. Map physical buttons to load tracks, jump cues, or trigger things over the API.
• Automation and home-lab glue. Home Assistant, scripts, a lighting or visualizer rig reading the beat and status stream.

What would make this easier

A few things I'd love thoughts on:

• Should I publish a proper API reference or an OpenAPI/JSON schema, so a frontend author (native or otherwise) has a contract to build against rather than reading the handlers?
• How stable does the API need to be before people would build on it? I'm happy to commit to not breaking things casually, but I'd want to know what matters.
• Auth and cross-origin / cross-device: today it's a trusted-LAN tool. A native app on your laptop talking to a daemon on a booth Pi is exactly the case worth getting right early.
• Would a tiny client library (a thin wrapper for a language or two) lower the barrier, or is the raw API enough?

Two questions for anyone reading:

1. What frontend would you actually want, or build? A native GTK/Qt app? Something else?
2. Is there anything missing from the current API that would block it?

If this resonates, I'm glad to document the API properly and treat frontends, native ones included, as a first-class way to extend Vynull.

[kayrozen]

I do love a browser client but i think that's a great idea to publish an openapi schema. Maybe also dockerfiles so the daemon could be run anywhere.

[vynulldev]

I drafted #58 to add the OpenAPI schema/spec, and a few more features to that end like SwaggerUI for playing with the API, CORS support, and now SSE streaming for events. The work you did adding browser streaming support is a huge win and opens the door to a lot more neat things we can do with frontends!

[kayrozen]

sooooo, i was inspired by the artcraft thing and i am testing a platform agnostic rust client (packages) and webassembly. I also am helping with the Pajoniiir project (ESP32-P4 player with usb support), i am testing with both sharing the same UI mockups i made.

[vynulldev]

That's awesome! Yeah it's honestly amazing how far we've come just in the last couple months even. Pajoniiir looks neat I'll check out your fork, I don't have an ESP32 or a DDJ controller though.

I think there's a lot of potential to make something great here. Let me know how I can help!