unav-devs team mailing list archive

[Joerg Berroth <joergberroth@xxxxxx>]

Hi!
>>>> 1. The "analyzing speed cameras for longer trips takes annoying long
>>>> time with the app not responding to any input.
> 
> I saw it 2 months ago and I improved to show the message.
> It's so complicate to fix that because there so many speed camera POIs
> and we need to filter who of them are in the route.
The bbox you are using might not be the best approach, see below.
If we can keep this box smaller we will have less pois to analyze.
> In my tests, 600km of route, analizing them should be ~30 seconds. I
> think in Germany there are more speed cameras :P
Exactly, so it takes up to 1-2min (if you are crossing almost complete
Germany, dep also on network connectivity ofcourse).
> I'll think on it again, but I'm afraid that it's not a good solution
> about this :(
It is. I saw If we build a poly-box with better dimensions by e.g. first
checking the maximum expansion of the route. So a bit more pre- but much
less post-processing. i will think about it.
> 
> 
>>>> --> Might it be possible to send this function into background !?
> It could be a possible solution, but a independent javascript thread
> would be so complicate of debug if we have a problem in the future.
Yes.
> 
> 
> 
> How many time took the check of speed cameras in your case?
> 
> 
> 
>>>>     if not in the js, maybe we can do it on qml level and send the
>>>> results
>>>> back to the webview.
> 
> 
> 
> Avoid QML <> HTML calls as we can ;) They are not good friends :P
I don't see this as that critical. More critical for example is memory
handling of oxide ;-) ...
> 
> 
>>>> 2. There are states, where the app those not return a gps signal anymore.
>>>> This seems not to be related to the gps-service, as restarting the app
>>>> leads to a clear signal again.
> 
> 
> Probably it is this bug or the next I'm telling you:
> https://bugs.launchpad.net/unav/+bug/1562783
Yes, it's not a freeze but the loss of GPS. That's why I was thinking
the process treatment on our side is not that perfect.
But the more I test the more random it is...
i will play more with it.