Sync actor private data between messaging and helper loop inside actor core VI

[norbi597@gmail.com]

Hello Everyone,

In my application, I must have a deterministic loop (or loops) with precise timing and I want to use the actor framework to change the behavior of that loop (or loops) by messages. I have multiple hardware and I want them to have the same functionality, so I created a base actor for them, and a child actor for each hardware. There is a helper loop (could be a timed loop) in the ancestor class Actor Core VI. In that loop, there are dynamic dispatch VIs. My problem is that Labview copies the class's private data in the message handler (generic actor core), so the helper loop will work on a different set of data. I want to sync the data between the message handler and the helper loop, without knowing what data the child actors have.

Here is the Actor Core VI:

Here is the class hierarchy:

 The Actor Core is overridden only in the "PS" class.

The problem is very similar to this discussion's problem. The difference is that I must use that class wire synced with the messages, because of the protected dynamic dispatch VI inside my helper loop.

What do you think, what is the best way to implement this? 

Thanks in advance,

Norbert

[cbutcher]

You might have some success with a Time-Delayed Send Message, and a public/community scope class VI which calls your other (dynamic dispatch, protected scope) VIs.

If you make the message class private within the library of the actor, then it won't be sendable from outside.

Community scope on your wrapper VI can be used to prevent calling it from something other than the message if you believe that would be an issue.

[drjdpowell]

A common actor-oriented design pattern is to have a helper loop send periodic messages to the main loop.  The helper loop manages timing, but the actual action is done by the main loop, using all the data the main loop has access to.  You might consider that.

BTW, the design you show does not have "precise timing" at all, and will experience "drift" over time.  Using an ordinary "Wait" node is never precise.

[Hypiz]

But this pattern is also potencially not precise because of other messages coming to the actor which is in some cases unacceptable.

[norbi597@gmail.com]

I can agree that the ordinary "Wait" is not so precise, anyway, that's not the point right now 🙂 

A wanted to avoid messaging for that particular method because messaging is a non-deterministic process and can increase jitter.

Apart from that, I will stick to messaging, with a higher priority. 

Thanks!

[drjdpowell]

Computers are non-deterministic unless you are using an RT system, with jitter of, at very best, plus minus 1000 microseconds.  Your actor can handle a lot of short messages without increasing that jitter.

[norbi597@gmail.com]

I forgot to say that this application will run on a cRIO system. Anyway, messaging using private message classes works fine. I realized that hiding the functions was more important for me than precise timing (the loop frequency is relatively low).

Thank you all!

[Hypiz]

It's not as rare as you obviously expect that all the messages are "short" (e.g. Open DB Connection) and the same pays for the jitter.

Would you replace a critical blue helper loop on cRIO? But this is an OT so let's it be.

[drjdpowell]

---

@Hypiz wrote:

It's not as rare as you obviously expect that all the messages are "short" (e.g. Open DB Connection) and the same pays for the jitter.

Would you replace a critical blue helper loop on cRIO? 

---

It's rare to have long running actions that are independent of the helper loop.  You can't save data to the database in the helper loop if the connection is being opened.  If OP's "PS" is a power supply, then "Configure Power Supply" probably takes a long time, but you can't use the Power Supply in the helper loop while it is being configured.  

You can have reason for a time-critical helper loop, of course, but in that case OP's design is not sufficient.