Hello, is there any plan for oxlint to support custom plugins?

[rendeyuwei]

I saw on the official website that there were mentions of future support for custom plugins. I’d like to ask if there’s a planned timeline for when this might begin? I noticed this plan was listed quite some time ago. Our company project is currently using oxlint as a replacement for eslint, and this feature is quite important for us. Thank you!

[overlookmotel]

I'm afraid we have no timeline at present. It's something we're discussing at the moment.

We're aware this is something which would be of great value to the community, and we very much would like to support it. But please bear in mind it's a large undertaking, especially finding a way to support custom plugins in JS without destroying the performance which is one of the selling points of Oxlint. We think we may have the beginnings of a solution, but it's quite experimental.

So, in short, sorry no nothing solid we can offer at present. But, if it's any consolation, it is very much on our minds!

[nrayburn-tech]

@overlookmotel has there been any discussion for custom plugins in Rust? I’ve only ever seen JavaScript plugins discussed (which is understandable given that is what the community has available today for ESLint).

[overlookmotel]

Not to my knowledge. As you say, JS plugins is the more popular request. And, as far as I'm aware, native plugin systems are quite tricky, because of all the possible target platforms.

We're open to any popular plugin being implemented in Oxlint itself, so that's usually the easiest way.

I'd be very interested to know what your use case is for custom Rust plugins, if you're able to say? Something company/domain-specific?

[nrayburn-tech]

I don’t have a use case on hand, I just wasn’t sure whether it was ever discussed/considered and if so what the outcome of that was.

The only thing that comes to mind is the more that is implemented in Oxlint adds more burden to the Oxc team which is obviously limited. It doesn’t seem like a sustainable approach, but that depends on how the team size (and external contributions) scales over time. This probably isn’t something to worry about right now.

[overlookmotel]

I'm not aware of there having been any serious discussion of this.

Some tools support WASM plugins, but that's quite a lot of work too. Especially as our AST is still somewhat in flux, so then you have to deal with versioning.

Personally, I think probably we can find a way of supporting JS plugins and making them run almost at native speed (as long as the plugin isn't hugely complicated). We think we have the beginnings of a solution to achieve that, and I'll be prototyping it over next few months. If we can manage that, it would be a bit of a "have cake and eat it" situation, and make the case for native plugins even less compelling.

The only thing that comes to mind is the more that is implemented in Oxlint adds more burden to the Oxc team

Yes, totally agree - this is the problem. We will never be able to implement every plugin that everyone needs. There's likely also a "long tail" of niche plugins, which are not commonly used. But what probably is common is to need some plugin which isn't "vanilla".

But our current direction is to try to fulfill that need via plugins written in JS.