Weird diagnostics behavior, seems to be out of sync

[JRasmusBm]

Hi!

Thank you so much for this library and all the work you do with CoC. It has really made all my workflows feel so much more modern!

I am experiencing some weird behavior with the diagnostics when using coc-python, and wonder if you have seen something like it before. This is how I experience the problem:

Whenever I make a change in a file and save it the analytics give false errors. They refer to errors that are on other lines or about my variable names being different than how I typed them. The invalid diagnostics remain when I save the file with :w, but go away when I source it again with :e.

Here's a gif showing the problem.

Is this a known issue or am I doing something wrong?

Best Regards

Rasmus

[JRasmusBm]

My coc-settings.json (Confirmed that problem persists with downloadChannel: "stable")

{
  "diagnostic.errorSign": "✗",
  "python.formatting.blackArgs": ["--line-length", "80"],
  "python.formatting.provider": "black",
  "python.analysis.downloadChannel": "beta",
  "python.workspaceSymbols.rebuildOnFileSave": true,
  "python.jediEnabled": false,
  "rust.all_features": true,
  "suggest.autoTrigger": "always",
  "suggest.noselect": false
}

[itsfarseen]

This occurs in Haskell (both hie and hls) and Rust (coc-rust-analyzer) too. In Rust, it happens almost always when I save file. Doing :e fixes it till the next :w.

[itsfarseen]

Just did a :PlugUpdate, problem seem to be gone in my Rust project.