Experimental Physics and
Industrial Control System

"Hartman, Steven M." <[email protected]> · Mon, 2 Apr 2018 17:57:27 +0000

> On Apr 2, 2018, at 12:28 PM, renato sanhueza <[email protected]> wrote:
> 
> Maybe I didn’t express myself correctly, but I was asking for the scenario in which the client doesn´t know a priori the details about the server that it wants to monitor (e.g. it might be a thermometer, a power supply, a fan, an antenna, etc).

Hi Renato—

Knowing only the name of the variable, the client can request the structure. For example, using the command-line pvget utility . . .

$ pvget BL6:Det:Metadata
BL6:Det:Metadata
structure 
    time_t timeStamp 2018-04-02T13:32:08.794 19815342
    epics:nt/NTScalar:1.0 proton_charge
        double value 0
    epics:nt/NTScalarArray:1.0 time_of_flight
        uint[] value []
    epics:nt/NTScalarArray:1.0 pixel
        uint[] value []

 . . . tells me that the variable BL6:Det:Metadata has a structure consisting of a timeStamp, a NTScalar value for proton_charge, and NTScalarArrays for time_of_flight and pixel. 

The client can then decide it wants to monitor only the timeStamp . . .

$ pvget -m BL6:Det:Metadata -r timeStamp
BL6:Det:Metadata
structure 
    time_t timeStamp 2018-04-02T13:38:55.189 19839726

BL6:Det:Metadata
structure 
    time_t timeStamp 2018-04-02T13:38:55.206 19839727

BL6:Det:Metadata
structure 
    time_t timeStamp 2018-04-02T13:38:55.223 19839728

> So the question is:
>  
> Can a client use metadata to understand a message from an unknown server? (known as dynamic communication) 
> Or the software developer has to known the structure of the server to make the client (static communication)  

If you are using Normative Types, the client will understand the structure of the data. (In the example above, the structure is application specific, but the structure is built from standard normative types.)

However, using your example, for the client to understand that a structure is intended to represent a fan and that the speed is the variable to control, would likely require additional information on the client side. 

-- 
Steven Hartman
[email protected]

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System