Feature request: Support formulas in the Write PV action value field #3941

Activity

1. shroffk commented on Sep 29, 2026

   

   shroffk

   on Sep 29, 2026

   Member

   More actions

   This is broadly a good idea.

   • Ideally, I think there should be some indication of what is going to be written.
   • should the value only be calculate when the user presses the write button? ( in this case it might be more difficult to address the previous point cause the value is not know until it is written )

   I would like to add this as a codeathon project.
2. shroffk

   added

   discussion

   codathon

   on Sep 29, 2026
3. kasemir commented on Sep 29, 2026

   

   kasemir

   on Sep 29, 2026

   Collaborator

   More actions

   In another view, the GUI shouldn't "do" anything. The GUI shouldn't try to figure out what an appropriate setpoint might be.
   Move that to the IOC. In this case, add something like this:

   record(ao, "CopySetpoint1")
   {
     field(OMSL, "closed_loop")
     field(DOL, "CalculatedSetpoint1")
     field(OUT, "Setpoint1 PP")
   }
   

   .. and in the GUI have the button write 1 to CopySetpoint1.PROC.
   If you need more than just a "copy", use the "calcout" record to compute the appropriate setpoint.

   That way, you can use EDM or MEDM or any other EPICS display tool, or caput from the command line.

   👍React with 👍2Reacted by Kunal Shroff and Georg Weiss
4. shroffk commented on Sep 29, 2026

   

   shroffk

   on Sep 29, 2026

   Member

   More actions

   Order of preferred solutions best to worse

   Doing it in the IOC ( as Kay showed ) > Doing it via formula functions > via some script ( what you are trying to avoid )

   👍React with 👍2Reacted by Kay Kasemir and Georg Weiss
5. rosstitmarsh commented on Sep 29, 2026

   

   rosstitmarsh

   on Sep 29, 2026

   ContributorAuthor

   More actions

   In my example the calculations are done by an IOC on a diagnostic device, I don't want the Phoebus screen to "do" anything. The GUI is just providing a simple mechanism to apply the value from one IOC to another. I could implement it with an extra record as shown but I prefer to keep my diagnostics IOCs as simple as possible. It could also be done by the magnet IOC at the other end but to keep it simple the magnet doesn't need to know that a particular diagnostic exists. To me the formula way is the simplest solution, this isn't business logic, just a nicety to save operators copying a value or making mistakes.
6. georgweiss commented on Sep 30, 2026

   

   georgweiss

   on Sep 30, 2026

   Last edited by georgweiss

   Collaborator

   More actions

   ...apply the value from one IOC to another.

   Isn't that a classic EPICS pattern? I'd second @kasemir here.

   ...keep my diagnostics IOCs as simple as possible

   Wouldn't that put potential complications on the front end (=Phoebus)?

   👍React with 👍1Reacted by Kay Kasemir
7. kasemir commented on Sep 30, 2026

   

   kasemir

   on Sep 30, 2026

   Last edited by kasemir

   Collaborator

   More actions

   As for magnet vs. diagnostics IOC, I assume your magnet IOC has the "Setpoint" and the diag IOC has the "CalculatedSetpoint".
   You now have at least 3 options:

   1. Place the "CopySetpoint" record in the magnet IOC so it can "pull" the calculated setpoint from the diag IOC
   2. Place the "CopySetpoint" record in the diagnostics IOC so it can "push" the calculated setpoint to the magnet IOC
   3. Place the "CopySetpoint" record in a new "automation" or "helper" IOC so it can "copy" the calculated setpoint from the diag IOC to the magnet IOC

   There are valid arguments for each option. EPICS really doesn't care. You can simply move that "Copy" record between IOCs, and database links will magically morph into network links.