Open p3r7 opened 4 years ago
Found the culprit!
It seems that I introduced native-complete in my setup.
It seems that ^[*^U echo ^Y
gets (sometimes?) suffixed to the input commands (see native-complete-style-suffix-alist
).
This suffix sometimes get sent to the interpreter, stucking it.
This must be native-complete/company doing its thing, for proposing completions back to the user.
This explains why the issue occurs when typing commands but not when pasting them.
native-complete
works fine when I spawn remote shell wo/ using with-shell-interpreter
.
I can kinda prevent the issue by commenting out this line in with-shell-interpreter
. This disables setting explicit-shell-file-name
(which is in most cases redundant with setting shell-file-name
as both gets looked up by most shell fn).
After checking out, native-complete
doesn't seem to rely directly on this var!
So this must have to do with some lower-level comint/shell fn being called an crapping out silently.
I have a strange issue w/ my setup.
When spawning a remote interactive shell w/
friendly-shell
orfriendly-remote-shell
and sending a cli command to the spawned shell, it sometimes hangs.This seem to only happen with commands split in several words (e.g.
apt search toto
) and is not happening 100% of the time. Even more strange, it doesn't seem to happen when we copy/paste the command in the buffer but only when we type it (letter by letter).It seems to hang when sending the command and not at the output processing / display stage.
^C
allows to cancel the command and retrying it makes it work.~~The problem seems to occur starting with Emacs 26.2 and is also present on 27.2. I'm not 100% positive about when things started to break. More precisely I don't know if it's an Emacs/TRAMP regression, a
friendly-shell
/with-shell-interpreter
regression or due to another package I introduced.~~ (see next message)