Hi,

> why not offer it as a default via /dev/gps?

> So the only useful solution for desktop linux seems to write a
> QGpsdPositionInfoSource. This can then easily be made system default as
> there's no additional configuration required to connect to a local
> gpsd.

That assumes that the user has gpsd installed and that it's the preferred 
method of abstraction to use for accessing GPS on linux systems.  

At present the positioning part of the Location API is only supported on 
desktop platforms via QNmeaPositionInfoSource, since the data could be coming 
from /dev/gps, from a Bluetooth device, or from pretty much anywhere.  If there 
was a standard one-size-fits-all-approach for a particular desktop platform 
then we'd probably have either considered it or at least asked the community if 
anyone wanted it badly enough to contribute a backend.  QNmeaPositionInfoSource 
is provided for those circumstances.

I believe that the developers of the Maemo 5 and Maemo 6 backends had access to 
gpsd and went with different solutions, so gpsd doesn't seem to be 100% the 
right choice (although some of those other solutions make use of gpsd).

There's work being done on a geoclue based backend for MeeGo and we have IP 
based positioning methods for the desktop on our radar, so it's still an area 
of active work.  At some point we'll need to look at what Linux desktops / use 
cases fall in between being able to use the MeeGo backend and being able to get 
away with using an IP based position source.  

I'm a little skeptical that there's a clear, single solution and/or a 
significant amount of demand for that case but it's entirely possible that I'm 
dead wrong on either or both fronts :)

Cheers,

Dave

> Am Sonntag 03 Oktober 2010 schrieb [email protected]:
> > You need a custom QGeoPositionInfoSource. Since your source seems to
> be an NMEA source you might want to give QNmeaPositionInfoSource a try.
> Ok, i have looked a little deeper into this. The main reason that
> GNmeaPostionInfoSource is not available as a default for
> QGeoPositionInfoSource may be the fact that the Nmea source needs to be
> told where to get its data from 
> 
> There may be cases where having an nmea class may seem useful,
> especially on embedded devices where any additional layer costs
> precious memory. But in general there's imho no valid reason for qt to
> talk directly to an nmea device. This is like giving qt its own video
> card driver.
> 
> So the only useful solution for desktop linux seems to write a
> QGpsdPositionInfoSource. This can then easily be made system default as
> there's no additional configuration required to connect to a local
> gpsd.
> 
> Has anyone actually started working on something like this? If not i
> might put some work on this, but i'd really like to see these efforts
> not no be a waste of time since someone wither better qt knownledge
> does the same.
> 
> Till
> _______________________________________________
> Qt-mobility-feedback mailing list
> [email protected]
> http://lists.trolltech.com/mailman/listinfo/qt-mobility-feedback

_______________________________________________
Qt-mobility-feedback mailing list
[email protected]
http://lists.trolltech.com/mailman/listinfo/qt-mobility-feedback

Reply via email to