Request for advice for a way forward for resolving Self and self

[dillonhicks]

This isn't rebased on the current changes that support "path tree" and is I have crate and super working reliably but Self and self are problematic.

So I need help tracking down these two problems:

• ResolvingSelf works in the simplest cases but fails in nested impl contexts, see: https://github.com/rune-rs/rune/compare/master...dillonhicks:keyword-paths-advice-pr?expand=1#diff-18418503cbb8b7076324353e3c9c592a . It looks like it indexes the inner function Foo::outer_1::$block0::Bar::inner_2 but the later query produces no meta. Any pointers?

• Resolving self fails if not prefixed with :: because it is parsed as a Expr::Self_. Should I just remove this in favor of parsing "paths"?

[udoprog]

Resolving self fails if not prefixed with :: because it is parsed as a Expr::Self_. Should I just remove this in favor of parsing "paths"?

Yeah. Definitely.

Resolving Self works in the simplest cases but fails in nested impl contexts, see: [..]

It should be possible to resolve Self from impl_items for now:

rune/crates/rune/src/indexing/index.rs

Line 44 in d8caf61

pub(crate) impl_items: Vec<Item>,

You'll have to store this in Query and look it up through an opaque identifier. Similar to how it's done to figure out where a path is resolved from today:

rune/crates/rune/src/indexing/index.rs

Line 1070 in d8caf61

path.id = self.query.insert_item(self.items.item());

It follows that of Self is encountered outside of an impl block it should cause an error.

[dillonhicks]

I rebased onto 8a43abb. I'm going to give this one last attempt before I admit defeat.

I am going to focus on Self and ignore self at the moment since that might be a bit of a distraction because of the parsing changes.

Thanks!