When I have an Erlang process ID, using pid/3 or list_to_pid/1 functions (which internally do the same thing) I can get the process for debugging purposes.
Process = pid(0,4,1).
Process = list_to_pid("<0.4.1>").
So the question is; what about ports?
There are lots of function interfaces which accept both process() and port() data types, like register/2. So I need to know if there is a way to get ports by their ID (e.g. #Port<0.567>) like processes. Is it prohibited? If so, is there any reason for it?
I'm not aware of anything in the standard library to help with this, but there is the recon_lib:term_to_port/1 function in the recon library that does what you want. For example:
1> {ok,L} = gen_tcp:listen(0, []).
{ok,#Port<0.687>}
2> L = recon_lib:term_to_port("#Port<0.687>").
#Port<0.687>
This code opens a listen socket on an ephemeral port and stores it in variable L. It then asserts, using pattern matching, that the result of passing the stringified port #Port<0.687> to recon_lib:term_to_port/1 returns exactly the same socket as L.