command: fix signalling to command process group

[adambabik]

This change also fixes a regression that broken running interactive commands, like runme in runme, run from the CLI in da2c38c.

It also disables FIFO which should mitigate #635 but a proper fix is needed.

Relates to #636

[adambabik]

@sourishkrout I was able to reproduce the issue and verify this fix using grpcurl.

[adambabik]

@sourishkrout there is a genuine problem with TestRunnerServiceServerExecute_WithStop so getting this PR back to draft.

I dug into this and when doing CTRL-C in the terminal, you actually send a SIGINT to the whole process group (a great thread that explains it thoroughly). Now I remember why I implemented this :) However, when running a server and executing commands by it, the server and the commands share the same process group and we kill the server when we send a signal to the process group; the effect is the reported bug.

It seems like the ideal solution is to make the command its own process group, just as it is done by bash. I will try that. In fact, there is code doing that but it's not used in v2.

[sourishkrout]

I dug into this and when doing CTRL-C in the terminal, you actually send a SIGINT to the whole process group (a great thread that explains it thoroughly). Now I remember why I implemented this :) However, when running a server and executing commands by it, the server and the commands share the same process group and we kill the server when we send a signal to the process group; the effect is the reported bug.

That explains why the kernel server suddenly dies 😆.

[sonarcloud]

Quality Gate passed

Issues
3 New issues
0 Accepted issues

Measures
0 Security Hotspots
80.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

[adambabik]

@sourishkrout ready to test and review.

[sourishkrout]

👍 LGTM