Labview RT: Share TCP connection refnum between parallel loops

[TomD3]

Hi,

I am programming a RT application that sends data via TCP/IP from the RT to a computer. Now, it might happen that the computer crashes and the TCP connection is lost. In my RT program, I would like to have sort of a "TCP connection monitor" that listens constantly for a TCP connection. Once a new TCP connection is established, a second, parallel loop should  use this "new" refnum for sending data. (There will always be just one client.)

Now I wonder about the best practice to implement this. I thought of using the same approach as in the example "TCP named service", i.e. creating a TCP listener and constantly waiting for an incoming connection. However, I cannot store that connection refnum id in a local variable that could be used from within another loop...

Why is that? What would be a nice way of distributing the connection refnum from one loop to the next? I would prefer to avoid STM since the amount of data might become large and perfomance might decrease...

Thanks

[rolfk]

Can you explain more why you can't pass a refnum through a local variable? There is nothing in LabVIEW that makes this specifically impossible but there might be situations that can make this look more difficult than you expect.

[TomD3]

What I tried: I right click on the output terminal "listener ID out" of "TCP Wait on Listener". The submenu "Create" did not offer "Local Variable", as it does for e.g. front panel objects...

[Mark_Yedinak · Accepted solution]

All local variables are associated with a front panel control or indicator. You cannot simply create a  local variable from the output of a subVI.

[TomD3]

oh, I wasn't aware. Thanks!

Just on a side note: would you implement the whole thing differently, e.g. without local variables?

[crossrulz]

---

@TomD3 wrote:

Just on a side note: would you implement the whole thing differently, e.g. without local variables?

---

2 options:

1. Make an Action Engine (AE) that handles the communication.  The entire Send-Read Response (or just Send) should be handled inside of the AE.

2. Make a Queued Message Handler or Actor that runs in parallel with your loops.  You use a queue to command that loop to send data and possibly read a response.  That QMH should be in full control of the connection.