Experimental Physics and
Industrial Control System

"J. Lewis Muir" <[email protected]> · Tue, 11 Jan 2011 12:12:30 -0600

On 1/11/11 10:42 AM, Mitch McCuiston wrote:
>  Hi Lewis,
> 
> The reason I wanted to use annotations is so that I could use AOP to
> instrument my classes intead with CA code intead of repeatedly writing
> it for each PV backed object.  For example, if you wanted to implement a
> callback that would be executed when an PV monitor fired you would
> typically do the following in JCA/CAJ.
> 
> 1.  Create a context.
> 2.  Create a channel to the PV you want to monitor.
> 3.  Create the monitor and map it to the java method you would like to
> callback to.
> 4.  Convert PV types to Java types when necessary.
> 5.  Execute pendIO calls when necessary.
> 6.  Write code to tear down the channel once you're done.
> 
> I feel like this is a manual process that could be automated for the
> majority of cases.  If you just define the mapping of the callback
> method to the PV you are monitoring (with annotations in my case) you
> can use AOP to instrument your class with the rest of the code.  The
> monitor case is almost identical to a Message Driven Bean/POJO in the
> J2EE world except your using a PV name instead of a JMS destination
> name.  I also like the fact that my Java abstractions are decoupled from
> the CA code, which makes things like JUnit testing much easier.

Hi, Mitch.

OK, I see what you're trying to do.

I still prefer regular Java code.  For example, with a good EpicsPv
implementation, I could do something like this to monitor an EPICS PV's
value:

  EpicsClient client = new EpicsClient();
  EpicsPv ringCurrent = client.newPv("ioc23:RingCurrent");
  ringCurrent.monitored();
  ringCurrent.addPropertyChangeListener("value",
    new PropertyChangeListener() {
      public void propertyChange(PropertyChangeEvent event) {
        System.out.println("Ring current: " + event.getNewValue());
      }
    });

I think in your annotations approach, this would be:

  @PV(name="ioc23:RingCurrent", accessType=PVAccessType.MONITOR)
  public void printRingCurrent(double value) {
    System.out.println("Ring current: " + value);
  }

This is a simple example, but to me the problem is that you're moving
your code into annotations which makes it harder to understand what's
going on and makes it less flexible.

It's harder to understand because I have to look up the PV annotation
implementation to figure out how it works.

It's less flexible because I can't change the annotations at run-time.
What happens if I want to stop monitoring the ring current PV in the
above example?  In the no-annotations approach, I just call
"ringCurrent.unmonitored()".  In the annotations approach, how would I
do this?

Still, your annotations approach might be easier to use in some
situations.  If I just want to set up a "callback" or do a simple get or
put, I just know, "When the PV changes, this annotated method gets
called with the new value," or, "When I call this method, it will set or
get the PV value," and it just works.

Lewis

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System