Re: tramp (2.5.0-pre feature/native-comp/2e25eebfbd25b131b6d0fcff4e60f7a8773d912b); Tramp hangs when executive start-file-process-shell command in non-dired buffer (python-mode)

[Michael Albinus]

Michael Lingelbach <m.j.lbach@gmail.com> writes:

> Hi Michael,

Hi Michael,

> Sorry about that; I've attached the compressed logs. They are rather
> large (50 mb), the only two commands I did were the two below, one
> command sent to dired, the other to a python buffer.
>
>> I'm trying to get a remote languag server running via tramp. When I
>> attempt to execute any remote shell command in a dired buffer, such as:
>>
>> ```
>> (with-current-buffer "box-physics"
>>   (start-file-process-shell-command "mspyls-remote" "*mspyls-remote*" "stty" 
>> "raw" ";" "python-language-server" "2>/tmp/mspyls-remote-16-stderr"))
>> #<process mspyls-remote>
>> ```
>>
>> The process executes.
>>
>> When I repeat the process in a python file:
>> ```
>> ELISP> (with-current-buffer "train.py"
>>   (start-file-process-shell-command "mspyls-remote" "*mspyls-remote*" "stty" 
>> "raw" ";" "python-language-server" "2>/tmp/mspyls-remote-16-stderr"))
```
>>
>> The process hangs emacs indefinitely until interrupted,returning
>> ```
>> *** Eval error ***  Couldnt find local shell prompt for /bin/sh
>> ```

Well, in the traces I see the two asynchronous calls:

--8<---------------cut here---------------start------------->8---
23:06:01.389431 tramp-send-command (6) # cd 
/home/mjlbach/Repositories/lab/interactive_physics/ &&  exec 
<<'ef1d49f95bfc199388bc0a3a8c09e181'  env 
PS1\=/ssh\:jedi-wireguard\:/home/mjlbach/Repositories/lab/interactive_physics/\ 
\#\$\  /bin/sh
(
stty raw ; python-language-server 2>/tmp/mspyls-remote-16-stderr
) </dev/tty
ef1d49f95bfc199388bc0a3a8c09e181
23:06:28.399649 tramp-process-sentinel (5) # Sentinel called: €#<process 
mspyls-remote>€ €exited abnormally with code 143
€
--8<---------------cut here---------------end--------------->8---

This process has run for 27 seconds, and then it was aborted. Tramp
doesn't know the reason.

--8<---------------cut here---------------start------------->8---
23:06:57.590647 tramp-send-command (6) # cd 
/home/mjlbach/Repositories/lab/interactive_physics/ &&  exec 
<<'ef1d49f95bfc199388bc0a3a8c09e181'  env 
PS1\=/ssh\:jedi-wireguard\:/home/mjlbach/Repositories/lab/interactive_physics/\ 
\#\$\  /bin/sh
(
stty raw ; python-language-server 2>/tmp/mspyls-remote-16-stderr
) </dev/tty
ef1d49f95bfc199388bc0a3a8c09e181
23:07:20.962226 tramp-process-sentinel (5) # Sentinel called: €#<process 
mspyls-remote>€ €exited abnormally with code 143
€
23:07:48.104411 tramp-process-sentinel (5) # Sentinel called: €#<process 
mspyls-remote>€ €killed
€
23:08:48.471035 tramp-process-sentinel (5) # Sentinel called: €#<process 
mspyls-remote>€ €killed
€
--8<---------------cut here---------------end--------------->8---

Here we have two sentinel calls; I guess the second and third ones are
from your kill command.

Tramp has no control over asynchronous processes, once they have been
started. Maybe you check the error file /tmp/mspyls-remote-16-stderr?

You can also try the calls manually from a shell on the remote machine,
like

--8<---------------cut here---------------start------------->8---
cd /home/mjlbach/Repositories/lab/interactive_physics/ &&  exec 
<<'ef1d49f95bfc199388bc0a3a8c09e181'  env 
PS1\=/ssh\:jedi-wireguard\:/home/mjlbach/Repositories/lab/interactive_physics/\ 
\#\$\  /bin/sh
(
stty raw ; python-language-server 2>/tmp/mspyls-remote-16-stderr
) </dev/tty
ef1d49f95bfc199388bc0a3a8c09e181
--8<---------------cut here---------------end--------------->8---


If this doesn't help, you might rerun the test, with tramp-verbose set
to 10. This will tell us much more details, but it will also9 be much
larger. Maybe you send it off-list to me only, in case.

When you run the tests again, pls give the processes a different name,
in order to distinguish them in the traces.

> Best,
> Michael

Best regards, Michael.