Fix invalid trait generation

[GuillaumeGomez]

Fixes#33919.

r? @steveklabnik

And a screenshot to come along:

[eefriedman]

fn eq(&self, other: RefCell<T>) -> bool
fn ne(&self, other: RefCell) -> bool

That doesn't look quite right.

[GuillaumeGomez]

Oh indeed. Going back to add missing information then.

[GuillaumeGomez]

New screenshot (lifetimes and type parameters are now generated as well):

[GuillaumeGomez]

A more complete screenshot to see lifetimes as well:

[steveklabnik]

I think conceptually, I like this change, but I am gonna review the code more a bit later. If anyone else is happy with the code, feel free to r+ it.

[steveklabnik]

This seems okay to me, but I am still getting to know rustdoc. @rust-lang/tools do any of you have thoughts on this patch?

[alexcrichton]

The issue this is fixing, #33919, is mostly just a dupe of #14072 which is a general problem that rustdoc doesn't substitute type parameters, so it just shows up this way. In general I think this may be a pretty ad-hoc solution which doesn't actually get us much farther.

It looks like this only handles Self, right? And it also only handles &Self and and Self as well? That is, I don't think cases like Box<Self> seem to be handled.

I think this would be best done as a preparation pass during rustdoc, either during the "clean" phase or as a dedicated pass in rustdoc itself. The rendered is already pretty complicated with lots of functions taking lots of arguments, so offloading as much as we can elsewhere seems beneficial.

[GuillaumeGomez]

@alexcrichton: That's what I tried at first but got quickly stuck with my lack of knowledge on the "low" code. I'll try to redo it in the "preparation pass".

[bors]

☔ The latest upstream changes (presumably #34541) made this pull request unmergeable. Please resolve the merge conflicts.

[steveklabnik]

@GuillaumeGomez are you still working on this?

[steveklabnik]

Closing due to a lack of activity, let me know if you want to keep up with this!

[GuillaumeGomez]

Ah damn, didn't see the first comment. I'll, but don't know when.