Nathan Ridge

> This feature is already supported in Microsoft's C/C++ extension Do you know how their feature works at the LSP level? Does it use `textDocument/documentHighlight`, or something custom?

The part about base declarations makes sense to me. Not sure about the overloads part: * Are you thinking specifically about cases where multiple overloads are declared one after the...

FWIW, I'm more skeptical about the second case (overloads in general), since a doc comment for one overload may talk about parameters that don't exist for other overloads, describe semantics...

I guess this is a consequence of https://github.com/clangd/clangd/issues/1258

> First I just wanted to confirm this is still the expected behavior? If yes, don't you think this can be improved in this case? (either on server or vscode...

> Sure, here is a basic [repo](https://github.com/MNASTM/clangd-issue-682) to give a try Thanks, this is very helpful. I can confirm that I see the same behaviour that you describe in the...

Ok, I've debugged this a bit, and I believe I now understand how the observed behaviour arises. As some background, clangd's symbol index has two parts: 1. the **dynamic index**,...

So, to circle back to the original questions: > this is still the expected behavior? If yes, don't you think this can be improved in this case? The choice to...

> So first if my understanding is right we can claim current behavior is related to `clangd` server, not `vscode-clangd` client: do you confirm ? Yep. > Considering, yes, some...

> I'm aware about https://github.com/eclipse-cdt-cloud/vscode-clangd so I' m getting to get an extra try based on. Possibly it may help but to be confirmed (I'll come back to you) because...