Experimental Physics and
Industrial Control System

Marty Kraimer <[email protected]> · Wed, 11 Feb 1998 09:47:16 -0600

I would just like to add a little more to what Bill said.

What I will address is only EPICS base and not the extensions.
All testing will be based on epics 3.13 and on vxWorks 5.3 (tornado).

For base I can think of the following things to check:

tsLib - The time stamp conversion library (Used by both base and
extensions).
drvTS - The code that obtains time stamps on iocs.

In addition base uses time related services from other sources.

CVS - The code management system.
      Although this does not affect releases if it breaks it could cause
problems.

Hosts - tsLib calls host routines to obtain the current host time. 

vxWorks supplied libraries. We are probably in good shape but must test.

NTP - Network Time Protocal. drvTS is a NRTP client.
      The current drvTS code is NOT prepared for the NTP timestamp
overflowing in 2036.

GPS - Many system directly or indirectly get their time from a system
that uses GPS.

If anyone can think of other time related things that could affect EPICS
base please send a message to tech-talk.

I have already asked William Lupton some quewstions about NTP, UTC
(Universial Time Coordinated), and GPS. His answers are helping me
understand. I still have a way to go.

My goals are:

1) Set up an isolated network with a solaris host and an ioc.
   Then set the time to various times starting Jan 1, 2000 and verify
that epics works.
2) Add a chapter to the Application Developer's guide that clearly (or
at least try) to
   explain how the ioc gets and processes time.

Any help, advice,etc is appreciated. However, remember that I will
concentrate on epics base.

Marty Kraimer

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System