[time-nuts] question about GPS time

[David Bridgham]

On 3/28/24 2:03 PM, Tom Van Baak via time-nuts wrote:

> If a second needs to be added to UTC it is named 23:59:60


This brings to mind a question I've been meaning to ask one of these 
days.  I guess today is a good day.

I was contemplating building a system that would keep its internal time 
in TAI.  That makes more sense to me than jumping the internal time 
around to keep up with leap seconds; that conversion, if needed, can be 
done in a UI library.

In support of this, I was looking over GPS receiver data-sheets to see 
what I had to work with.  The GPSes I found all liked to report times in 
UTC, rather than TAI or GPS time.  Hmmm.  But one thing I noticed was 
that the seconds field of that UTC time report was defined to be 
[00..59].  Uhhh, so what do they do when a leap second comes along?  
Does the GPS receiver double up with :59 or maybe it rolls over to :00 
and doubles that one up?  Neither of those are very good answers.  Or 
maybe the documentation is just wrong and the receiver actually does the 
right thing and report :60.  I suppose documenting it wrong is better 
than doing it wrong.

So while I'm curious about that, my real question is whether there's a 
way to get GPS time out of a GPS receiver.  Or, lacking that, is there a 
way to reliably get the information out of a GPS receiver as to what 
leap-second offset it's currently using to calculate the UTC that it's 
reporting.  Yeah, I *should* be able to figure that out with my own 
leap-second database but how can I be sure that the GPS is really using 
the same list of leap-seconds that I have?  Better if it just tells me.  
Better still if I could get GPS time without the leap-second offset applied.

Thanks,
Dave