RE: PV name changes and epics extensions

["Rock, Judith E." <jrock@slac.stanford.edu>]

Hi Emmanuel, 

You asked about SLAC - we're using the PV part of IRMIS at SLAC.  Our rdb is Oracle, and we're running the PV crawler, along with client crawlers for alh, Channel Watcher, Channel Archiver, and some sequence programs.  Please let me know if you'd like more info.

cheers,

Judy

---

From: tech-talk-bounces@aps.anl.gov [mailto:tech-talk-bounces@aps.anl.gov] On Behalf Of Don Dohan
Sent: Monday, August 27, 2007 8:45 AM
To: Emmanuel Mayssat
Cc: epics
Subject: Re: PV name changes and epics extensions

Emmanuel

First, IRMIS consists of various parts - PVs, hardware and cabling.
This discussion only concerns the IRMIS PV part.

Emmanuel Mayssat wrote:

IRMIS is definitely one of the solution I am looking for.
But how difficult is it to implement?
  

You have to install a relational database management system, and then make the appropriate PERL crawlers work (usually this is an NFS issue: providing network access for IRMIS to the .db and.dbd files that the IOC uses to reboot).
It really helps to have a standard ioc build structure, to help the PV crawler find the associated resource files.

is IRMIS for Oracle databases only?
  

We use MySQL at APS, SNS and SLAC use Oracle, the Canadian Light Source uses Microsoft Sequel Server,  for example.  I think that there is also some work on PostGreSQL

Are simple crawlers easy to program?
  

It depends on the domain.  The pv crawler is the most complicated since it is trying to emulate the IOC boot process.  In spite of this complexity, we have had valuable contributions/fixes to the PV crawler code.
For PV client crawlers, again it depends on how sophisticated the client handles nested macro substitution, etc
A lot of these crawlers have also been written.

It will be an interesting challenge for the hardware part of IRMIS to handle the plug-and-play aspects (that you were discussing in another thread). That is another topic.

Don

is there any other solution out there?

Is anybody on this list at SLAC and using IRMIS?
I definitely would like to have a short demo in the near future.

Thanks,
--
Emmanuel


On Fri, 2007-08-24 at 14:08 -0500, Don Dohan wrote:
  

Emmanuel

This is one of the features of the IRMIS relational database. 

IRMIS consists of, in part, all of the PVs in your system.  A pv crawler 
(a perl program) detects when an IOC reboots, then scans and expands 
that IOC's st.cmd file to load the IRMIS database with that IOCs PVs.  
This scanning is unobtrusive to the IOC.

Other crawlers have been written to scan CA clients.  At the APS, we 
crawl for MEDM, ALH, and SDDS clients, and we are testing the SNL 
crawler written at SLAC.  We intend to crawl save/restore files as well 
(future).  For each CA client instance, a list of resolved PVs is 
generated, and inserted into the IRMIS database.

Taking the MEDM client as an example, the crawler resolves the macro 
substitutions for nested MEDM screens, so that it tells you which .adl 
file in the chain needs to be fixed if a given PV is removed.

Crawlers for other CA clients have been written at other facilities - 
EDM, DM/2K for example.

Don Dohan

Emmanuel Mayssat wrote:
    

Another question about software upgrade...

Let's say that I change the name of PV in my epics database.
Potentially with this change, I could break epics extensions.
Medm of course, but also feedback system, PV backup, etc.

Is there a way to know which extensions will break if I change a PV
name? (may keeping track for each extension of the used PVs in a flat
file is the solution, but is there any other, smarter ones?)

--
Emmanuel