Adjust CodeActionResponse

[Xanewok]

Looking at textDocument/codeAction it seems to be returning heterogeneous array (Command | CodeAction)[] | null rather than Command[] | CodeAction[] | null so I adjusted that accordingly.

Discovered this while working on rust-lang/vscode-rust#609 - right now RLS sends only Vec<Command> but to fix the issue we also need send some CodeActions as well, hence why we need the heterogenous array.

r? @Marwes

[Xanewok]

A "weekend" project idea: one of these days it'd be great to have a bindgen but for TS which would just scan the .ts file and spit out the Rust bindings, maybe using https://github.com/FreeMasen/RESS? @matklad interested? 😝

[matklad]

Yeah, I was thinking writing a simple ts parser for that :D

The problem I see, though, is that there's really no single source of truth for the protocol definition :(

The markdown spec does not contain various experimental features, and simply extracting typescript snippets from md would be a bit of the pain.

Also, I mildly hate inconsistent naming in the protocol (some things are X, some things are XRequest), and I would probably prefer to rename them on the library level.

So, I think the pre-requsite for that work is to compile a machine-readable definition of the protocol first, separate from the md speck

[Marwes]

👍 Released as 0.58

[Marwes]

A "weekend" project idea: one of these days it'd be great to have a bindgen but for TS which would just scan the .ts file and spit out the Rust bindings, maybe using https://github.com/FreeMasen/RESS? @matklad interested? 😝

Would be even better if the LSP spec used something like JSON schema microsoft/language-server-protocol#67 . The debug server protocol does actually provide that which let me generate the types for that protocol https://crates.io/crates/debugserver-types https://crates.io/crates/schemafy