Flatten highlights

[matklad]

The highlight function currently returns a Vec<HighlightedRange>. However, those highlighting ranges can be nested. I believe that for something like #[cfg(feature = "foo")] we produce the range for the whole attribute, and then, separately, a range for a string literal.

It seems like all the clients of syntax highlighting actually want a flat list of highlights. We should refactor highlight so that it produces such flat list. One obvious way to do this is to post-process the current return type, but, given that internally we walk the tree in dfs manner, perhaps we can just not produce nested things in the first place?

The relevant code is here:

https://github.com/rust-analyzer/rust-analyzer/blob/a5da660bd5953c046db5f0bff854c00d830da9c8/crates/ra_ide/src/syntax_highlighting.rs#L34-L38

[ltentrup]

I have had a look at the specific code and I would like to work on it.
My first idea is to use a stack Vec<Vec<HighlightedRange>> mirroring the DFS traversal. When leaving a node, one can implement the flattening.
To implement the flattening itself, one has to clone HighlightedRange but it does not implement Clone:

https://github.com/rust-analyzer/rust-analyzer/blob/a5da660bd5953c046db5f0bff854c00d830da9c8/crates/ra_ide/src/syntax_highlighting.rs#L27-L32

I am not sure about the semantics of binding_hash, can it be cloned or must it be unique?

[lnicola]

@matklad do we still want the rainbow stuff after #3820? It's only used in a test.

[matklad]

We need to keep it internally, as this is a useful (and somewhat tricky with respect to shadowing) feature to have. However, the impl should probably be changed to a separate request, independent of the main highlighting.

…

On Thu, 2 Apr 2020 at 14:07, Laurențiu Nicola ***@***.***> wrote: @matklad <https://github.com/matklad> do we still want the rainbow stuff after #3820 <#3820>? It's only used in a test. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#3447 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANB3MYPS7BRJHSWEBFWJOLRKR5ZHANCNFSM4LBHTOAQ> .

[kjeremy]

Did #3826 solve this?