Experimental Physics and
Industrial Control System

Marty Kraimer <[email protected]> · Fri, 30 Sep 2005 06:48:29 -0500

Jeff,

Please look at:

1) The power point presentation I am giving at the EPICS meeting about

the V4 database.

2) The wiki titled "V4 Design: Record Processing"

In particular note that

1) I hope that, to the user, there is no difference between database and

channel access links.

2) input, output, and process links can optionally request
   a) The record be processed
   b) notification when the record completes processing.

NOTE: Both documents use the terms wait and block. These names are

confusing and may change but to what?

This, of course, requires support from Channel Access.

Iattached the power point presentation in previous messages but will

attach it again.

Also something like 3) below is my guess about what CA will need.

Marty

Jeff Hill wrote:

Hi,

I am thinking about process passive and maximize severity, and where that
fits in the CA API.

Correct me if I am wrong, but my understanding is that in a database context
these are attributes of the link and not attributes of the individual IO
requests on the link. Therefore, if I made them into properties of the CA
request, and they changed from one request to the next that could be quite
vexing to the database - interested in your thoughts on that statement.

If the above statement is correct then it seems that we have some leading
candidates:
1) Include the PP and MS attributes as syntax standardized extensions to the
channel name and just let the database handle this.
2) Make PP and MS be writable properties of the channel and the database
would need to allow these parameters to change during the duration of the CA

link.

3) Pass a property catalog to the create channel call which specifies

attributes of the channel to set up. We already need the channel priority as
a parameter to this call so this would be a quite flexible mechanism. This
PropertyCatalog could be forwarded with the protocol element establishing
the channel with the server.

Option (3) is starting to look like a strong candidate coming at the expense
of complicating the API somewhat, but interested in your thoughts on the
matter.

Jeff
__________________________________________________________
Jeffrey O. Hill               Mail         [email protected]
LANL MS H820                  Voice        505 665 1831
Los Alamos NM 87545 USA       Fax          505 665 5107

EPICS_V4_Database.ppt

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System