Experimental Physics and
Industrial Control System

"Jeff Hill" <[email protected]> · Wed, 27 Aug 2008 14:33:08 -0600

> 
> Is it possible to extend channel access to take two optional parameters
> (for array fields only), so that we can read/write only a few consecutive
> elements of the array (N elements starting at index I, default to the
whole
> array)?

This _is_ on the list but it is currently somewhat further down the list.

> 
> Another idea, is it possible to extend the registering of monitoring
> callbacks to take some optional parameters?  And the IOC would check
> against those extra conditions to decide whom to send the callback.
> 

I am currently working on a Channel Access server upgrade that will allow
clients to specify generalized filter expressions, as character strings,
when they enter a subscription. The upgrade will also allow posting of
events containing service specific payload content. The boolean filter
expression will run against any named parameter in the payload as interfaced
by the Data Access library, and the subscription update is sent to the
client only when the expression evaluates true.

I spent the majority of my time on another project last year, but this has
recently changed and I now expend a much larger (majority) percentage of my
efforts on the server upgrade.

> 
> How much work will this be?  

Concerning your first request, this requires a protocol change. The first
decision will need to be whether this upgrade should be implemented as an
extension to the existing protocol or in the redesigned protocol (there is a
design document for the new protocol on the wiki if you are interested).

Concerning your second request, this is the top item on my Channel Access
upgrade list. There is quite a bit of work to do, but I am making good
progress. When I finish the new server there will be additional work related
to defining new services and or extensions to the execution semantics for
the database.

> Will it break all the existing clients?

No. When we transition to a redesigned protocol, we will maintain code
interfacing with the old protocol during the interim.

Jeff

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System