The "folks" with this particular determination were just one: Damien Miller. He's the "djm" in the mailing list history you linked in post #12. He is still the lead developer for OpenSSH.
It's possible Damien may be willing to revisit the issue, if there is some reason the ControlMaster option cannot be used. You could always ask him.
I don't see the caveats as critical, because they are related to TCP session performance and master session control. Neither should be an issue if you enable ControlMaster and set unique ControlSockets for each background shell you wish to manage.
Another alternative is using a getppid(2) syscall from the background shell, as I'd originally envisioned you needed.
|