Experimental Physics and
Industrial Control System

"Allison, Stephanie" <[email protected]> · Tue, 16 Dec 2003 08:56:24 -0800

Hi Steve,

> I noticed your question regrading linking of MCA on linux which you posted
> to tech-talk in October 2002.  Did you have any success with that?

We compiled mca on Solaris and created a shareable library.  I've not yet compiled on Linux as nobody has requested it.  I don't think the compiled mca library is used much.  In my current project (SPEAR), we use Solaris matlab for RF only.  All other matlab apps are run on Windoze (and use mca).  And for RF, the Solaris compiled matlab executables are currently using the "unix" command with "caget" and "caput" for get/put PVs.  They haven't been converted to mca yet, if ever.  One worry I have about mca is the hardcoded 1 second timeout on all CA calls (perhaps this is changed in a newer version).  This may be a problem for some of the RF matlab apps.  I know a user in the NLCTA project was having a problem with the 1 second timeout in mca when she was forced to go through a slow gateway process.

> We're currently looking a Matlab with the idea of using it as part of our
> control system for the Australian Synchrotron, and I'm evaluating it on a
> Red Hat linux box at the moment.  I've successfully compiled MCA (in
> particular mcamain.cpp) and have a little example application up and going
> which monitors one process variable. I've found it to be very 
> unstable... I very frequently get segmentation faults when I run the 
> application and, after some debugging, the problem appears to be within MCA itself.
> 
> I don't suppose you've run into anything like this in your investigations?
> My next step is to try it out on Windows and see if I get the same
> instability.

No.  The very very little bit of testing I did on Solaris didn't show this problem.  

Stephanie

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System