Experimental Physics and
Industrial Control System

Marty Kraimer <[email protected]> · Tue, 27 Aug 2002 07:07:34 -0500

Does your application call gettimeofday to get the time stamp?

If so it uses seconds since the UNIX epoc (1970) and microseconds within the
second. Another, less likely, possibility is that your application calls
clock_gettime, which uses seconds since the POSIX epoch (1970) and nanoseconds
within the second. Another possibility is that your application calls time which
just returns seconds since 1970.

For EPICS call tsLocalTime for 3.13 or for 3.14 call epicsTimeGetCurrent.
Marty Kraimer
Kevin Tsubota wrote:
Hi All,
I have a channel access application that we inherited from someone else a

long time ago that is
returning a timestamp of seconds since 1/1/1970. This application also has

its own camonitor

utility that returns the current date. However when I use the base

camonitor I get timestamps 20yrs

in the future.
Upon further investigation I found comments that EPICS uses 1/1/1990 as its

epoch. So I'm guessing

here that EPICS camonitor is taking the timestamp from my application and

adding 20yrs. Is this

correct or am I _way_ out in left field? 
Thanks in advance.
Kevin
Code snippet from EPICS R3.13.0Beta12/base/include/tsDefs.h
/*--------------------------------------------------------------------------
-
-
* TS_STAMP
*
*    This is the form taken by time stamps in GTACS, and is the form used
*    by the various tools used for dealing with time stamps.
*
*    The time stamp represents the number of nanoseconds past 0000 Jan 1,
199
0,
*    GMT (or UTC, if you prefer).
*
*---------------------------------------------------------------------------
-
*/
typedef struct {
    epicsUInt32    secPastEpoch;   /* seconds since 0000 Jan 1, 1990 */
    epicsUInt32    nsec;           /* nanoseconds within second */
} TS_STAMP;

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System