Background: I use eglot with c-mode and every time I use c-fill-paragraph, eglot breaks—it reports a syntax error when there isn’t. It seems that eglot’s image of the buffer’s content (or what it thinks is in the buffer) is corrupted. I can fix this by revert-buffer. I have a feeling that this has something to do with after-change-functions because that’s the only way eglot gets updated with buffer changes. So I setup an experiment. It goes like this:

Open track.el and eval-buffer; then go to test.c and M-: (mytrack); then open the tracking buffer “trackbuf” in another window. Now if you edit in test.c, the change should be reflected in trackbuf. Now try M-q. Both buffer ends up in a corrupted state. 

I tried to look at c-fill-paragraph but it’s too complicated. I hope someone with more expertise in c-mode could give some insight.

Yuan