Yo Ferry!
On Fri, 03 Jan 2014 21:25:03 +0100
Ferry Huberts <address@hidden> wrote:
Seems like a lot of work for zero gain. gpsd tries to keep the
servers as simple as possible. If the client wants to group things
the client should do so.
Well, I have a use-case for it. My use-case involves low-bandwidth
networks on which every byte counts.
How low? How many bytes does this really save? A UDP header is 8
bytes, is this really noticeable in your application? Is it worth
the added compleixty and latency?
---> [22:01:34] -- $GPRMC,.....
---> [22:01:34] -- $GPVTG,.....
---> [22:01:34] -- $GPGGA,.....
---> [22:01:34] -- $GPGSA,.....
---> [22:01:34] -- $GPGSV,.....
---> [22:01:34] -- $GPGSV,.....
---> [22:01:34] -- $GPGSV,.....
---> [22:01:34] -- $GPGLL,.....
Also, this pack mode
communicates to a listener an as-complete as possible GPS state in a
single packet.
gpsd data goes stale really quickly. Every time in the past we
have attempted to accumulate state we have had to revert.
Furthermore, from my point of view gpsd is the server, gps2udp is a
simple client program of gpsd that performs a simple transformation
and sends out the result.
Yes, gps2udp is the client to gpsd, but it is the server to the
udp clients. And you are making the transformation less simple.
I'm trying very hard to do the right thing here and upstream work we
have.
Yes, and thank you for that. On our end we want to be sure changes
to git head suit all users. As you have seen, if you work with us
a little we often get to a good, if delayed, solution.
RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97701
address@hidden Tel:+1(541)382-8588