Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Wed, 20 Aug 2008 13:35:37 -0500

> >> I know I can use
> >> ==========================================
> >> dbLoadRecords ("db/test.db","device=${DEV},channel=1")
> >> dbLoadRecords ("db/test.db","device=${DEV},channel=2")
> >> dbLoadRecords ("db/test.db","device=${DEV},channel=3")
> >> dbLoadRecords ("db/test.db","device=${DEV},channel=4")
> >> ==========================================
> >> in st.cmd script and epics will get the system environment "DEV" so
> >> that I don't need to modify my ioc package and just use
> >> export DEV=blahblah
> >> on each device. But I kind of like my st.cmd script to be clean and
> >> neat, thus I made some .substitutions file like:
> >> ======================
> >> file "db/test.db" {
> >>     pattern { device, channel }
> >>         { "${DEV}", 1 }
> >>         { "${DEV}", 2 }
> >>         { "${DEV}", 3 }
> >>         { "${DEV}", 4 }
> >> }
> >> ======================

Since there really isn't much difference in content between these two, you can 
still achieve a clean and neat st.cmd file by moving all of the dbLoadRecords 
commands into a secondary script that you include from the main one:

== st.cmd ==================
# Load test.db for ${DEV}, channels 1 through 4
< loadTestDatabase.cmd
============================

== loadTestDatabase.cmd ====
dbLoadRecords ("db/test.db","device=${DEV},channel=1")
dbLoadRecords ("db/test.db","device=${DEV},channel=2")
dbLoadRecords ("db/test.db","device=${DEV},channel=3")
dbLoadRecords ("db/test.db","device=${DEV},channel=4")
============================

> I need testAll.db to be loaded many times so I don't like to put a lot of
> dbLoadRecords commands in the st.cmd script. Besides, I have like
> five or six macros in the .db file so that I can make it universal to the
> instruments in my charge. But when it appears in the st.cmd script, it's
> a long line and not supposed to be broken into several lines which,
> I'm sorry, doesn't seem perfect to me. While if I use the "pattern" form
> of a .substitutions template, the macros are just grouped logically.
> It's very nice except it can't accept the system environment.

My philosophy about this kind of question is that I really don't like taking 
initialization values or paths from environment variables, because it makes 
it harder to reproduce the same boot conditions at a later date.  All it 
takes is for someone to manually set a few environment variables before 
starting the IOC and you now can't come back three weeks later and get 
exactly the same result unless you remember exactly what those settings were 
â do you keep your ~/.login or ~/.profile files under CVS/SVN control?

In an operational control system reproduceability is important, so IMHO all 
initialization values should be given in startup scripts or .substitutions 
files that can be saved to a revision control system.  That's why I wouldn't 
*want* the IOC to expand environment variables in a .substitutions file 
(although I believe they do work in the filenames, because they get expanded 
in a different point of the code).

HTH,

- Andrew
-- 
Talk is cheap. Show me the code. -- Linus Torvalds

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System