Experimental Physics and
Industrial Control System

Ben Franksen <[email protected]> · Fri, 29 Oct 2010 01:08:56 +0200

Hi All

this is a poll to determine whether anyone is actually using the 
sequencer with backends other than Channel Access and so needs support 
for the pv layer between the sequencer and the backend data transfer 
protocol.

I would rather like to get rid of the pv layer for several reasons.

First, I have no idea how to get detailed information about KTL (Keck 
Task Library), the only other supported backend. I googled but all I 
found was some papers and a more than 10 years old web page. Continued 
support for a KTL backend is thus not something I can do myself. I 
don't even have any header files.

Second, the implementation of the CA backend adds serious overhead; for 
instance, it happily uses new/delete all over the place (at runtime), 
something which we try hard to avoid in the rest of the epics codebase.

Third, to fix certain deficiencies I *may* need to to use CA features 
that are not part of the pv interface. I cannot say anything definite 
at this point, as I am still in the process of exploring options.

Fourth, interfacing to other control systems is probably better done 
using gateways. For instance, at BESSY we have made good experiences 
with an OPC/CA gateway.

Fifth, I am not sure anyone is really using the pv layer other than for 
CA.

So, if I am wrong and this is a feature you are indeed still actively 
using, or if you think that there are other reasons to keep it, please 
speak up.

If I do not get any affirmative answers during, say, the next 3 weeks, I 
am going to abolish the pv layer in one of the next releases, how soon 
will depend on how desparately I need to go down to the CA library 
level.

Cheers
Ben

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System