Harald JOHNSEN wrote:

Another problem not related to pure performance is that the first retrieval of metar data can block FG for a long time (perhaps one minute) when the metar server is not accessible (or when there is any network problem). The code does a lot of (useless) retries. Perhaps its simpler not to update the metar on IO error and continue silently.

I have noticed this too. We are doing the metar fetching in a separate thread to avoid this specific problem, but I wonder if something in that process has broke, or if some how the metar thread can still manage to block the main thread while it is out fetching data. I haven't had time to poke around with this myself, but it would be worth trying to come up with a way to reproduce the problem so one of the developers could poke around and try to figure out what is going on. Thread code is tricky and there could be a subtle blocking bug in there some place.


Curtis Olson        http://www.flightgear.org/~curt
HumanFIRST Program  http://www.humanfirst.umn.edu/
FlightGear Project  http://www.flightgear.org
Unique text:        2f585eeea02e2c79d7b1d8c4963bae2d

Flightgear-devel mailing list

Reply via email to