Closed sarmong closed 3 months ago
Yes it's a problem with bob itself (maybe rust doesn't handle SIGUSR1). I will look into this but might take a while since Rust doesn't kill subprocesses with SIGTERM which will take some time to figure out unless I used a bit dep like nix crate.
Side note: for windows users there isn't really anything to do regarding this since windows TerminateProcess is not something that can be caught and handled.
@sarmong Could you test the PR?
Neovim still terminates, however now the terminal (or is it shell) is not messed up after termination
It is meant to terminate, so it works.
No, it is not. This is USR1 signal, not INT. Any other version version of nvim (from source, .deb, .appimage) doesn't terminate.
Oh, I thought the expected behavior was not the fucked up terminal which was caused by SIGKILL so I simply changed it to SIGTERM, is the expected behavior is to keep working basically?
Yes. Neovim can catch USR1 via Signal
autocommand, but it is not possible when opening bob version of neovim.
Updated the PR let me know if this is the behavior u expected. (cant really test this, since I am on windows)
Works properly now, thanks!
Neovim supports
Signal
autocommand. However, when the signal is sent to the neovim instance installed with bob, neovim terminates with 138.This doesn't happen when neovim is built from source, or via official appimage.
To reproduce:
bob use latest
(the same with nightly)kill -USR1 $(pgrep nvim)
Happens on Debian 12