Triggering flycheck-popup-tip clears the selected region

[hartikainen]

I confirm that...

• I have searched the issue tracker, documentation, FAQ, Discourse, and Google, in case this issue has already been reported/resolved.
• I have read "How to Debug Issues", and will use it to provide as much information about this issue as possible.
• The issue can be reproduced on the latest available commit of Doom.
• The issue can be reproduced on a stable release of Emacs, such as 27 or 28. (Doom does not support development builds like 29+ or any version ending in .50 or .9x)

Expected behavior

With eglot enabled, when I set mark and move the cursor within a buffer, I would expected the region to stay active and expand/shrink based on my cursors position. Here's an example of the expected behavior, without eglot, that is, otherwise the same setup as the current behavior below, but recorded after eglot-shutdown.

without-eglot.mov

Current behavior

With eglot enabled, when the cursor moves onto a line with eglot error, the region gets cleared/unselected. See the video below. The region selection works as expected on the first couple of lines, but when I move the cursor to a region where eglot reports an error, the region suddenly disappears.

with-eglot.mov

Steps to reproduce

1. Open emacs with (lsp +eglot) and (python +lsp +pyright +conda +cython) configurations in init.el.
2. Write a python file with syntax error, similar to the example in the videos above.
3. Open the python file in a new buffer and make sure that eglot is enabled.
4. Set mark in the buffer.
5. Move cursor over to a point where eglot reports an error.
6. The region gets cleared.

System Information

https://pastebin.com/EMjHtP6w

[hartikainen]

I want to note that there's a syntax error, i.e. a missing comma, on line 15 in the code above. Adding a comma makes the region work as expected. I think the bug still stands as I would expect the region to stay active despite the syntax error.

[totallyuniquelily]

I've run into what I believe to be this issue, and it can be triggered without eglot (though if it turns out to just be similar I'll open a new issue).
Posting this in the hope that a possibly more minimal repro is useful.

I have eglot disabled, but I lose my region in elisp scratch buffers when I move my cursor over an error.
Disabling flycheck-mode fixes it for me (though that might just be because there's no errors)

I've managed to reproduce with only the following:

1. disabled all modules aside from emacs-lisp, syntax, and (default +bindings +smartparens)) and doom{,-dashboard,-quit}
2. Open a new scratch buffer, and set emacs-lisp-mode
3. Write code. It can be correct, as flycheck will whine about there being no heading or footer.
4. Same as in original report, set mark and move over an error.
   You may also press shift-down while on a line with an error, and that line will not be selected

2024-04-07.22-07-11.mp4

my doom info:

• with stripped down modules (as listed above): https://pastebin.com/s9YAG9vv
• with my usual config (just in case) https://pastebin.com/AAPFUrTh

[hlissner]

This is probably flycheck-popup-tip's doing. See if enabling the +childframe flag on the :checkers syntax module fixes this for you (it'll use a child frame to display those errors instead).

EDIT: sure enough: flycheck/flycheck-popup-tip#14. And this will only affect non-evil users, thanks to this advice.

[totallyuniquelily]

can indeed confirm that +childframe fixes it!

[hlissner]

As of 6756911 this should be resolved for folks not using +childframe. Thanks for bringing it to my attention!