RE: PV aliases

["Thompson, David H." <thompsondh@ornl.gov>]

One  <possible> application for being able to add aliases after iocInit
might be an app where a bunch of calc records are banked in a soft IOC
ready to be used by just setting the INP fields, the CALC, field, and of
course the name (alias). Could be done on the fly.


-----Original Message-----
From: tech-talk-bounces@aps.anl.gov
[mailto:tech-talk-bounces@aps.anl.gov] On Behalf Of Andrew Johnson
Sent: Monday, October 08, 2007 11:50 AM
To: Heinrich du Toit
Cc: tech-talk@aps.anl.gov
Subject: Re: PV aliases

Heinrich du Toit wrote:
> Why not put the aliasing in a PV gateway?
> 
> This way you don't have to change the IOC's
> And you don't have to change and client sofware.
> You simply need to change the software on one gateway.

The disadvantage of using a gateway for aliasing is that you now have 
the relevant PV names listed in two places, so if you do eventually 
change the name of the original PV (because everyone is now using the 
alias) you have to remove it from the gateway configuration as well as 
change it in the IOC, and these changes have to be synchronized.

You might also want to avoid running large arrays or quickly changing 
signals through a PV gateway for performance reasons; there is no 
performance effect of doing aliasing on the IOC (other than to the PV 
name lookup process).  Access Security could also be a problem through a

PV gateway.

> On Mon, 2007-10-08 at 10:01 -0400, Lawrence T. Hoff wrote:
>> 	I may have been the one who initiated the
>> discussion a few years ago. I encountered a situation
>> which I believed was best addressed using server-side
>> aliases. I created a patch to EPICS core (which could
>> be loaded into a VxWorks system as a separate module),
>> which implemented server-side aliases.

See http://www.aps.anl.gov/epics/tech-talk/2003/msg00707.php for Larry's

original posting, although I do not recommend using this code verbatim.

>> P.S. As near as I could tell, there was not universal
>> enthusiasm for server-side aliases at that time. Maybe
>> at next week's EPICS collaboration meeting the discussion
>> could begin anew?

PV Aliasing has been on my "To Do" list for Base for quite a while (my 
boss wants it too), prompted by your demonstration that it's not hard to

do with the current data structures.  The important question arises as 
to whether we need the ability to add aliases after iocInit() - I would 
guess not often, but I can see circumstances where it would be useful.

The only issues remaining that I can see would be a review of database 
locking (if we allow addition of aliases to a running IOC) and Access 
Security issues, and to decide how to specify the aliases to the IOC.  I

would prefer to extend the DB file syntax rather than use something like

Larry's addPvAlias(name, alias) command, but adding alias support to 
dbLoadRecords() might preclude runtime aliasing.

I hope we'll be able to make some progress on implementing aliasing 
(amongst other things) at the EPICS Codeathon next April.

- Andrew
-- 
When a distinguished but elderly scientist states that something is
possible, he is almost certainly right.  When he states that something
is impossible, he is very probably wrong.  -- Arthur C. Clarke