How to correctly set up eglot and clangd formatting on Windows?

[bradJM]

Hi everyone,

I'm not completely sure if this is an eglot issue or a problem with my configuration (leaning towards the latter, though). I'm using eglot with clangd on Windows, and I have a .clang-format file with UseCRLF set to true. Emacs reports my existing files as having DOS line endings, and no control characters show up in my buffers. After I use eglot-format-buffer, all my lines end with a visible ^M character and the newlines are duplicated (ie an extra empty line between every existing line) if I check the file in another editor. I'm admittedly not very familiar with properly setting up Emacs different coding variables, but does this sound familiar to anyone?

My init.el contains:

(prefer-coding-system 'utf-8)
(set-default-coding-systems 'utf-8)
(set-terminal-coding-system 'utf-8)
(set-keyboard-coding-system 'utf-8)
(w32-set-system-coding-system 'utf-8)
(setq-default buffer-file-coding-system 'utf-8)

But honestly, that's just based on some Googling and trial and error. 😅 I've also tried using utf-8-dos explicitly with the same result. Thank you for any help!

[joaotavora]

Hello! It might be a bug in eglot, yes. To my understanding Emacs normally shows those characters of there's at least one non CRLF line. You have to setup a minimal reproducible example and then open an issue.

[bradJM]

Okay, thanks! I assumed I just had it configured wrong, but if it could be a bug then I'll get a repro together and make an issue. 🙂

[joaotavora]

It could be a bug, but I'd have to look at the event transcript, at least.

[joaotavora]

If you don't have all the elements to make an issue yet, don't. Paste an LSP event transcript here instead.

[bradJM]

I went ahead and made a quick reproduction of it just to make sure I could reproduce it with a stripped down config. I put the event transcript in a gist because it's quite long. The issue might be a bit clearer in this screenshot since the hidden characters are visible:

One thing I find interesting: what I thought was happening (the buffer was being rewritten with both CRLF and LF) is half true:

Before formatting:

$ file test.cpp
test.cpp: C source, ASCII text, with CRLF line terminators

and after:

$ file test.cpp
test.cpp: C source, ASCII text, with CRLF, CR line terminators