[gpsd-dev] New 0.5 draft of the SemPiTernal HOWTO

[Eric S. Raymond]

Frank Nicholas <frank at nicholasfamilycentral.com>:
> > The hardware management we keep should go into a separate refclockd that
> 
> Would refclockd be a part of NTP? 

Yes.

>If so, I’d be ok with that.  I just have no need for GPSd on a pure time server.

That implies a separate question, which is: should both refclockd and gpsd
parse NMEA0183?

I am the world's expert on parsing NMEA0183, expertise gained by a
decade of work on GPSD's driver and dealing with perversities you
could barely even imagine.  And I say "Hell, no" *shudder*.  I barely
even trust my own parser; I think I have excellent reason not to trust
anyone elses's.  There's just too much weird out there.

I am more sure that I am of what I had for breakfast that the refclock 20
parser has significant undiscovered bugs because NMEA parsers *always* do.
This is one of those cases where the best thing is to have just one
high-quality implementation and unit-test the hell out of it.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>