Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Mon, 26 Jul 1999 19:15:03 -0500

Bill Cruise wrote:
> 
> Well, I got quite a surprise when our system display showed
> 1 Jan 1900!  Turns out that tsStampToText() is used to break
> out the timestamp (which itself is OK), and it only has an
> option for a "YY" type format.  Unfortunately this function
> call is at the heart of our system.  We are running R3.13.beta11,
> in case there are version differences in this area.

tsStampToText actually supports two output formats:
    TS_TEXT_MONDDYYYY  Mon dd, yyyy hh:mm:ss.nano-secs
    TS_TEXT_MMDDYY     mm/dd/yy hh:mm:ss.nano-secs
Within the EPICS base source code the MMDDYY format is only ever requested
in test code or logging output from the sequencer.  Changing the
TS_TEXT_MMDDYY format to generate a 4-digit year would probably break more
code than it fixed due to the longer string generated, which the
application might not be able to handle properly.

The corresponding routime tsTextToStamp converts any 2-digit year < 90 to
a timestamp in the year 20yy (Note to EPICS maintainers in ~2080: you may
want to look at this) and handles 4-digit years correctly.

IMHO this is an application issue, not something that needs to be changed
in EPICS base.

- Andrew
-- 
I brake for tail-gaters

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System