Experimental Physics and
Industrial Control System

"Kasemir, Kay (ky9)" <[email protected]> · Wed, 20 Aug 2008 15:14:00 -0400

Hi:

In theory, most EDM widgets use the PV interface
defined in edm/lib/pv_factory.h.

They don't connect to an EPICS/Channel Access PV,
instead they use
   ProcessVariable *pv = the_PV_Factory->create("some_pv_name")
to get a pv.

Per default, or when you use "EPICS\\some_pv_name",
you get an EPICS Channel Access connection.
When you use "CALC\\..." or "LOC\\...", you get a different
type of PV, and in principle all you have to do is create a
ACNET PV Factory and hook it into this.
Allmost all the code you need to study would be in
edm/lib/pv_factory.h and edm/lib/pv_factory.cc,
so you can then estimate how hard that might be.

In practice, the ProcessVariable interface is somewhat
modeled after EPICS: A PV can be of type int/enum/double/string
or arrays thereof, it has a status and severity,
display and alarm ranges, units, ...
You'll have to shoehorn your ACNET data into this mold.
If your data doesn't have "precision", you'll have to make it up
or report "0". If your data has "color" or other properties
not in the ProcessVariable interface, it's unclear how to
get at it other than by using some special naming convention.

-Kay

[email protected] on 8/20/08 15:00 :

> Hi,
> 
> The E in EDM stands for extensible as I understand it.   There are two
> ways that I could think of a display manager(DM) being extensible:
> 1. extending the widget set
> 2. extending the set of PVs/data acquisition protocols supported by the DM
> My interest is doing the later without having to reimplement a whole new
> widget set as well.
> 
> We have users (mostly physicists and engineers but not programmers)
> doing development with EDM (and EPICS) that are unable to access useful
> machine parameters that are currently only available through ACNET.
> Here at FNAL we will probably never be an exclusively EPICS
> installation, because of the volume of machine parameters that are
> instrumented via home-grown ACNET.   I would like to extend EDM so that
> it can display legacy ACNET machine info in addition to EPICS machine
> info.   It would appear this would be possible at some level, since it
> appears EDM already supports EPICS and CALC PVs.   Is it
> reasonable/possible to extend EDM to support other "PV"s besides EPICS
> and CALC?   What is not clear so far to me is how much coupling there is
> between the widget set and the PV implementations.   Assuming I can
> develop an ACNET PV for EDM, would I also need to develop an ACNET
> widget set?
> 
> Thank you for your time.
> 
> Sincerely,
> Carl Schumann
> 
> 
> 

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System