Re: [gpsd-dev] Updated docs on NTP segment management

[Gary E. Miller]

Yo Eric!

On Tue, 3 Mar 2015 05:38:00 -0500
"Eric S. Raymond" <address@hidden> wrote:

> If, that is, we choose to do a redesign at all.  I think one of the
> measurement tasks going forward ought to be to profile both Pearly's
> JSON driver and the SHM driver against a large, representastive
> sample load.  

I'm trying that.  The last cut much better, but I still disagree with
many user visible design details.  The data looks better, if I can just
get it fudged right.

The precision calculation is all wrong.

The mingling of the two sources is problematic.

The configuration lines are idiosyncratic.

> If Pearly's driver doesn't add significant jitter over the SHM driver,

Worst case would be on like a RasPi.  Will try that soon.

> then I think it will make a much better base for forward development.
> JSON's extensibility makes it way, *way* superior unless
> deserialization overhead is a showstopper.

+1.

Pearly had a question about sending pps=on to an old gpsd.  I assume if an
older gpsd gets an unknown option it ignores it?  Would be good to 
know going forward that is a policy.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97701
        address@hidden  Tel:+1(541)382-8588