Experimental Physics and
Industrial Control System

"Jeff Hill" <[email protected]> · Wed, 3 Oct 2007 10:45:52 -0600

> This obviously implies recompilation of the extensions (medm, etc)
> against the new base, isn't it?
> In my case, my medm doesn't work correctly with 3.14.6+
> http://www.aps.anl.gov/epics/tech-talk/2007/msg00744.php
> Could that be related to channel access?
> 

I read that message, but unfortunately it's impossible to guess about the
cause of the medm freeze without more details. 

Hopefully other sites have experience using medm with some R3.14 release
greater than 6 (of the CA client library)?

> I am now able to make sense of the details of the CA client - CA server
> communication.
> 
> But then why do we have EPICS_CAS_BEACON_LIST default to the value of
> EPICS_CA_ADDR_LIST if the former is not set? 

This is done because:

1) It is necessary to remain backwards compatible with past behavior. In the
past we used only EPICS_CA_ADDR_LIST to configure the beacons, and didn't
use EPICS_CAS_BEACON_LIST.

2) If you are not specifying any ports in EPICS_CA_ADDR_LIST then it might
be convenient to configure both the search address list and the beacon
address list with EPICS_CA_ADDR_LIST.

> Can we send beacons and
> search requests to the same port? That doesn't make any sense to me...

If you need to specify non-default ports in EPICS_CA_ADDR_LIST then you must
use a non-default EPICS_CAS_BEACON_LIST.

In R3.15 I should add support for multicasting. That would make multi-subnet
configuration easier.

Jeff

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System