Re: [gpsd-users] Speeding up autobauding

[Alexander Carver]

On 10/23/2012 08:21, Eric S. Raymond wrote:
> Alexander Carver <address@hidden>:
> > > Can you test to see if the sync time drops at a lower speed?
> >
> > Yes, it does.  The autobaud almost always starts off at 4800 and
> > takes a few seconds to ramp up to 38400.
>
> And:
>
> >     Dropping the unused ones sped up the process.
>
> OK, that's what I was expecting to hear. Alas, there's no way of
> speeding this up in the general case - autobauding takes as long
> as it takes, because the real constraint is the serial I/O speed.
>
> >                                             I didn't know
> > about the fixed speed option or I would have used it because I keep
> > the one receiver in 38400 SiRF binary exclusively.
>
> That's why it's there.

Sync/autobaud speed wasn't much of an issue for me since the system 
isn't moving (so no need for rapid uptime to assist navigation) and 
really doesn't get restarted much at all.  The machine does one thing, 
keep time.  I'm still working out a few other features that will use the 
data from gpsd but since it's main purpose is my network clock, it does 
pretty well and a ten second autobaud isn't a huge deal.

If I recompile the newer version of gpsd then I'll remember the autobaud 
which will at least speed recovery along for times when the power goes 
out.  I need to see what the change logs say for the new version and 
then I still owe you a patch for DGPS status/satellite used.