fix: allow non_snake_case and dead_code lints to run within component functions

[yescallop]

Currently, the non_snake_case and dead_code lints will not run within component functions because the component macro emits allow attributes for the two lints, which allows room for users to make mistakes. This PR fixes the problem by converting the function name to snake case before prefixing it with __ and dropping the allow attributes for the two lints from the macro.

I have run cargo check on the library and all examples and see no additional warnings. We can have the CI confirm this.

[yescallop]

The server_fns_axum example failed to build because there are multiple component and server functions with the same name but in different case (rkyv_example and RkyvExample, for example). I changed component and server macros to emit different prefixes (__leptos_component_ and __leptos_server_) to address the problem.

[yescallop]

I don't know why, but I might've run into a cargo fmt bug touching seemingly unrelated code. I thought it would be better not to include those formatting changes, but apparently the CI would not be happy with that! So I included the changes and force-pushed again. Hopefully everything is OK now.

[rakshith-ravi]

Oh, nevermind, I see what happened here. Perhaps we can have a #[allow(non_snake_case, dead_code)] in the declaration, and another #[warn(non_snake_case, dead_code)] in the body to bring back the lint instead of changing the name of the fn or struct?

Not sure if that would work, but worth a shot

[rakshith-ravi]

Hi @yescallop. Wanted to check in to see if you had time to check out my comments?

[yescallop]

Hi @rakshith-ravi. I guess I get your point. Do you mean something like the following?

#[allow(non_snake_case)]
fn __Example() {
    #[warn(non_snake_case)]
    {
        // body wrapped, this will warn
        let A = ();
    }
}

But I notice that an inner attribute #![allow(non_snake_case)] on the module will not have an effect on the wrapped body, which is an adverse effect of this solution. What do you think?

[rakshith-ravi]

But I notice that an inner attribute #![allow(non_snake_case)] on the module will not have an effect on the wrapped body, which is an adverse effect of this solution. What do you think?

Oh my. Yeah I didn't think about that. Valid point indeed. It just seems very "hacky" to do a __leptos_component__{} 😅

Any other suggestion you have? If not, I guess this could do for now. We can always clean it up later if we think of a better solution

[gbj]

My only feedback would be that the server_fn_macro crate is used by/can be used by other framework, so perhaps simply __server_... and __component_... rather than __leptos_server etc. would be a better naming convention. Otherwise looks good to me.

[rakshith-ravi]

Changes made @gbj

[benwis]

@rakshith-ravi Looks like it's been a bit since this was made. If you can resolve the commits I'll see about merging it

[yescallop]

Conflicts resolved @benwis

[rakshith-ravi]

I've triggered the CI. Let's see if there's any issues

[benwis]

Thanks for the PR!