I've always wanted to do this!

I agree that it would be ideal if we can use the canonical parser implementation for this, but that it doesn't provide all the information needed today. There's some partial support for tracking code locations but the information doesn't make its way all the way through the compiler. I'd love to work on this but realistically I don't expect to have time anytime soon. If you'd like to take a shot at it, feel free.

@kentonv

Thank you for your reply.
I will first add byte positions to Node to reserve the positions of declaration.

struct Node {
  id @0 :Id;
...
  startByte @34 :UInt32;
  endByte @35 :UInt32;
}

Does this approach look correct to you?
Or should I add byte position to SourceInfo?
-> I decided to define it in SourceInfo according to Pull Request comment.

@kentonv

I have made a trial Pull Request for the changes to retain the position of declaration statements.
Could you please take a look?
#2126
As for obtaining the position of an Expression, I currently have no idea. It would be helpful if the information were available in the MemberInfo struct in node-translator.c++

@trickstar0301 Have you considered contributing to @norgor VS Code extension vscode-capnp which seems to be the most recent and feature rich?

@JonathanPlasse

Thank you. That’s really encouraging!

@kentonv
Thank you for merging the pull request.
I’ve been working on a Language Server implementation for Cap'n Proto, which you can find here: capnp-ls repository. It currently supports basic LSP features like definition jumping and symbol analysis.

The implementation relies heavily on kj for overall design. Its efficiency and simplicity made it an excellent fit for building this.

Would you be open to discussing whether this could become part of the official Cap'n Proto repository? I believe it could greatly enhance the developer experience for users working with Cap'n Proto schemas.

@norgor
Please take a look at sample client implementation. It demonstrates how to integrate with an LSP server.