I recreated this on my system and after thinking about this a bit, I believe PI Vision is functioning correctly.
I think the key point is that a day is 24 hours long, not 23 hours, 59 minutes, 59 seconds. So when you set the display limits as you do, you are actually missing a second of data from the day: between 11:59:59 PM and 12:00:00 AM.
As for the behavior in PI Vision, since you selected an interval that is less than a day, clicking the pane arrows causes the time interval to change by a difference of 23 hours, 59 minutes, 59 seconds rather than 24 hours. If you click multiple times, this becomes more obvious as the interval drifts away from midnight.
Please let me know if this makes sense.
1 of 1 people found this helpful
Exactly as Seth replied, the time period shown in bottom-center of the screen is derived from the set start and endtimes. This then also controls the shift when you scroll forward and back, so that's why you loose one second every time.
Now the misunderstanding is in the end time. This is "up to" and not "up to and including" (that's different across cultures around the world).
What you could do is to set your period to one day, and next set your end time to the start of the (next) day. Then the scrolling shifts one day at a time. Or set endtime "t+1d". Note that the starttime is always determined based on the time period, so if you change the starttime first, it will be changed when you change endtime or period.