Randy Lai

Ya, it will require some works to improve it.

Sounds like we should make sure to return 0 for NA column position at https://github.com/REditorSupport/languageserver/blob/master/R/document.R#L131

As a matter of fact, `code_point_to_unit` does return ~zero~ NA if the code point position is greater than the number of code units in the line. Though I am not...

Yes, it should be fixed. I am going to submit a new release to CRAN soon.

I think we need to at least support `eval = FALSE` by the chunk options.

Diagnostics are supported for Rmarkdown documents in `languageserver`'s perspective. @renkun-ken would know more about vscode.

It should just work.

You may try putting the path to R in settings.

try turning on the debug settings of LSP and R-IDE to see what error message you get?

It seems to me a locale issue. And the error occurred when running `R --slave -e "R.home('include')"` and I t gives a non ASCII char when it is not called...