Experimental Physics and
Industrial Control System

"Neidherr, Dennis Dr. via Tech-talk" <tech-talk at aps.anl.gov> · Mon, 26 Feb 2024 09:51:56 +0000

Dear all,

right now I am making the first steps to migrate our old control system to a combination of EPICS and python.
And because of the workload I have to do the migration stepwise and at some points I am forced to use specific interfaces / solutions which are already existing.

One of this solution, is that it seems I need a method to trigger the processing of a StreamDevice device in a python code, wait until the processing is done, and continue with the answer (and all that should be an atomic process, so no-one should interrupt the device in the meantime). Since this is a very basic process I would like to show you my solution and I would like to ask if you may have a better one (especially because I think my solution might have a "problem").

So the following python code shows how it might work:

import asyncio
from aioca import caget, caput, run

async def do_stuff():
    await caput('Variable.PROC', 1, wait=True, timeout=10)
    await asyncio.sleep(0.001)
    test = await caget('Variable.VAL')
    print(str(test))

run(do_stuff(), forever=True)

I tested this code and it already fulfills several requirements:
- the process on the (Stream-)device side is an atomic action. If a second sender tries to trigger a process, he has to wait and he does not get the result of the first one.
- it can be set up quite easy, only one EPICS record necessary for the whole process (because I start the trigger through the PROC field).

But there might be a drawback:
- it was only working since I included the "await asyncio.sleep(0.001)" in the code above. Otherwise I do not get a correct answer, I simply get the value from one cmd before. The reason might be that the "wait=True" in the caput-cmd waits only until the processing on the device side is finished and not until the result is already published. My problem is now: I am not sure, but this value might depend on the network and even 1ms might not be enough in all situations.

So as I already wrote: Does anyone have an idea how to solve this problem? Or should I go for a completely different solution (I am an EPICS beginner, so there might be a much smoother way I do not see yet).

Thanks a lot, and
Greetings Dennis

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System