Nathan Ridge

See also https://github.com/clangd/clangd/discussions/2001 for some related discussion.

> Is there a reason why `FastCheckFilter: None` does not allow for this potentially-slow deoptimization, since that is a `ClangTidy` option which is explicitly consenting to slow analysis? It would...

> I'd be curious to know what about the processing of the "preamble" would be so disproportionately intensive? But I suppose I don't fully understand what the preamble is, and...

> I am only interested in feeling confident about the diagnosis of the directly opened file. While ["whole-project" analysis](https://github.com/clangd/vscode-clangd/issues/158) would be a really neat option to selectively enable at times,...

Thanks for looking into this! Finally had a chance to read over the discussion in #2203 and here. First of all, I completely agree that the observed behaviour ("if I...

Regarding your call hierarchy example: > ``` > void a(int i, int j); > > void b(){ > a(3, 4) > } > ``` > > now having your cursor...

Finally, regarding your proposed fix: One category of cases where this would change behaviour is cases where a comma being used as a separator is claimed by some ancestor node,...

> Second, I am not sure I understand if the change to the LSP server accepting ranges would also fix the issue we have today where you have a "range...

FWIW, I came across this previous discussion of this topic: https://github.com/clangd/vscode-clangd/issues/307 and in particular the proposed mitigation idea from [this comment](https://github.com/clangd/vscode-clangd/issues/307#issuecomment-1069534638): > I think we should probably add an extension...

Confirmed that clangd is not sending a semantic token of kind `type` for the use of `__auto_type` on the second line of `main`, and it is for one on the...