Christian Clason

I think the issue is that the `textDocument/build` request does not take a `PositionParam` (only a `TextDocumentIdentifier`), so the followup `forwardSearch` command triggered from `build` doesn't have that information and...

Yes, sorry, my first comment was from before I did a bit of digging myself. You are right that (if I am correct) this requires a (minor) change in texlab...

Yeah, but at least for Neovim, I know how to (and will) do that ;)

This commit has introduced a regression with Neovim: if I set the root directory to `.` (corresponding to adding `.` as a `workspaceFolder`), I now get the following warning: ```...

That's a significant breaking change, then, as this was accepted before... In any case, conflating `.` and `/` doesn't seem correct to me, even if `.` is not valid --...

And if I pass `file://.` as URI, `lua-language-server` reports ```lua workspaceFolders = { { name = "file://.", uri = "file://file://." } }, ``` That can't be right :D

I see; that needs to be handled by Neovim then. Still, my point about `.` and `/` stands -- if the `name` must be absolute, that should not lead to...

Still, would it be too much to ask for to continue supporting relative paths? It worked fine before this commit, and getting absolute paths from a terminal editor started in...

Thank you, I appreciate that! I have brought up the issue with the other Neovim devs. Still, this does not fix my issue, since `.` is still incorrectly treated as...

yes, I am passing `.` as the `name`: ``` workspaceFolders = { { name = ".", uri = "file://." } }, ```