Example Does Not Work

I try to run the example. The second line of code workerToClient = parallel.pool.PollableDataQueue(Destination="any") gives me this error:

Error using parallel.pool.PollableDataQueue (line 28)

Too many input arguments.

I remove the Destination = "any" argument and that seems to get things working. But then I get to the for-loop in the example:

for idx = 1:5
    send(clientToWorker, [idx rand]);
    status = poll(workerToClient, inf)
end

and the computer locks up with no output. I force-quite (CTRL-C) and type

future.State

The result is 'finished'. It seems the process stopped and I never sent the stop message. Seems nothing I can do will keep the process running.

[Walter Roberson]

parallel.pool.PollableDataQueue(Destination="any")

has been supported only since R2025a.

[Walter Roberson]

You did not post enough context for us to be sure that clientToWorker and workerToClient have been created in ways compatible with the default destination.

[William]

I see the matlab provided example did not attach to this question. My appoligies. I'll copy it here:

Typed into the matlab command window:

pool = parpool("Threads",1);
workerToClient = parallel.pool.PollableDataQueue(Destination="any"); % Except that I had to remove the Destination="any"
clientToWorker = parallel.pool.PollableDataQueue(Destination="any"); % to get the error messages to go away.

In a seperate (processData.m) file:

function out = processData(workerToClient,clientToWorker)
out = 0;
send(workerToClient,"Ready to receive data.");
while true
    % Wait for a message
    data = poll(clientToWorker,Inf);
    if strcmp(data,"stop")
        send(workerToClient,"Stopped processing data on worker.")
        return
    else
        response = sprintf("Data %d received.",data(1));
        send(workerToClient,response);
        out = out+data(2);
        pause(1);
    end
end
end

Back to the command window:

future = parfeval(@processData,1,workerToClient,clientToWorker);
status = poll(workerToClient,inf)
for idx = 1:5
    send(clientToWorker,[idx rand]);
    status = poll(workerToClient,inf)
end

There's more code (to be typed at the command prompt) after this but matlab hangs at this point.

The full example code and instructions from Matlab is here www.mathworks.com/help/parallel-computing/send-messages-to-workers-using-pollable-data-queues.html

[Walter Roberson]

That example is marked "Since R2025a". It is not expected to work in R2024b.

[William]

Is there no way to do block a parallel process while it waits for data pre-2025a?

[Walter Roberson]

Sure, you can use poll() with an infinite timeout. https://www.mathworks.com/help/releases/R2024b/matlab/ref/parallel.pool.pollabledataqueue.poll.html

You have to be careful about what you are polling, as there is no Destination="any"