Alpha MwV2 vs MwV3

[seogrady]

Hey Oscar! I've been playing around with Rspc and I really like it. With the alpha version, how can I return a Result<> from a middleware function without having to use MwV3? I don't need to capture extra arguments in the middleware and can authenticate using headers/cookie. Are you able to provide an example?

[oscartbeaumont]

The difference between MwV2 & MwV3 isn't actually the arguments, it's just how the arguments are constrained. They both need the two arguments and the same function signature to be valid (The advantage of the unconstrained type is that you can provide non-closure types but this is mostly for super advanced stuff). The reason MwV2 exists comes down to this (playground) and I am still working on a coming up with a solution to this because it's really not great DX but Rust makes it very difficult.

Would be good for you to send through the code you've got cause I don't know if you're running on the latest crates.io release, main, or the v1-alpha. All of them handle stuff pretty differently. Can't wait till v1 is properly released but I just don't have the time right now to get it over the line.

[seogrady]

Thanks for the quick reply @oscartbeaumont ! Yeah it's a bit confusing with MwV2 & MwV3. I was having a look at the spacedrive usage of Rspc and saw they created a MwArgMapperMiddleware::<LibraryArgsLike>::new() that returns a Result<>. As there wasn't an example that returned a Result<>, I thought that MwV3 was required.

FYI – I checked out the repo and played around with the tests you have in src/alpha/mod.rs. Wrapping the result in Ok() compiles so 👍 . I think I just got a little lost looking through the code.

pub fn library<TLCtx>() -> impl MwV2<TLCtx, NewCtx = (TLCtx, i32)>
where
    TLCtx: Send + Sync + 'static,
{
    |mw, ctx| async move { Ok(mw.next((ctx, 42))) }
}

Cheers mate!