Re: [Flightgear-devel] Terrain sampling for elevation

[Adrian M. <kan...@gm...>]

On Saturday, December 01, 2012 19:19:14 geneb wrote:
> On Sat, 1 Dec 2012, Adrian Musceac wrote:
> > Hi,
> > 
> > Right now, due to some rather encouraging results I've had so far with my
> > little terrain sampling experiment, I've started a wiki page on this
> > topic: http://wiki.flightgear.org/Terrain_sampling
> > I've included a screenshot with a demo of three radio stations each about
> > 100 km away from the plain, and the relevant output from the performance
> > debug window.
> > The perfomance hit I'm seeing is quite similar to the scenery method,
> > except it's not as reliable, and the sqlite result can take sometimes
> > more time to be retrieved, sometimes less time, being thus quite
> > "jumpy".
> 
> This is a very cool feature Adrian.  Great job!
> 
> g.

Thanks! I'm still not quite happy with the performance of database queries.
Having a couple of navaids tuned in and the ocasional ATC transmission does 
not make much difference, but in a typical ground radar station scenario, 
there would be tens of aircraft that would need to be polled every couple of 
seconds. At the current state that would require too much cpu time. 

Still, I'm searching for ways to improve performance. The good news is that 
long range terrain sampling is possible even without having tiles loaded.
I will keep the wiki page updated and the code in my repo. Perhaps there's 
someone out there who has dealt with high performance sqlite queries before.

Cheers,
Adrian