[erlang-questions] Programatically Starting Erlang Nodes

[Dale Harvey]

How do people manage programatically starting and stopping seperate
erlang distributed nodes?

I would really prefer the option to do this inside erlang, making
starting a node as easy as node:start('hello@REDACTED'), test_server
start node uses a port under the hood which leaves the child node
closely tied to the parent node, starting -detached nodes has the
opposite problem of being very early to leave dangling nodes whenever
they arent wanted.

so the basic things I am looking for

easily create a node from within erlang
be able to to detach from child node and leave it running when parent node exits
a nice way to ensure the stopping child nodes actually shutdown, even
when erlang distribution falls down.

Can anyone think of elegant examples of setting up / tearing down
distributed erlang nodes? one managed outside erlang are also suitable

Cheers
Dale