[Xfce-i18n] genghiskhan at gmx.ca

[Harald Judt]

Am 19.07.2012 18:44, schrieb Genghis Khan:
> Hallo Harald,
>
> I have issues understanding these expressions:
>
> * Interval data applies to
> * Point data applies to
>
> May you please elaborate about them?

Despite not wanting to change anything before release, I decided it 
would be better to use wiser terms to avoid confusion.

So I hope the new translation strings explain the different types of 
data better and make the situation less confusing.

Sorry for regenerating PO files again.

The current situation regarding data and the task of the panel plugin is 
a bit difficult: For one part, the panel plugin is supposed to display 
the *current* weather. To be precise, its description claims so, and 
that's the way it worked in older versions. I have tried to make it do 
so as much as possible.

However, what it gets from met.no is *forecast* data. So there may be 
situations when the panel scrollbox shows 100% cloudiness (which can be 
verified as correct by looking out of the window in real life ;-)), 
while the weather symbol is happily claiming "Look it's so sunny 
outside." In reality, the weather symbol and precipitations are 
calculated for the next 1-3 hours and temperature, cloudiness etc. are 
provided for the current (or next) hour.

That's because the nearest point in time is chosen to get the values for 
temperature and similar values, which in many cases equals the start of 
the weather symbol time interval. So, at the start of the interval maybe 
it's not sunny, but at the end of it it probably will be. Of course, the 
end of the interval could be chosen for cloudiness, so that the weather 
symbol and cloudiness value are consistent. But then it would be a 
forecast for the cloudiness in e.g. 3 hours, and not the cloudiness at 
the current time (or as close as possible).

Trying to explain it in other words, the weather symbol tells you the 
change of the weather for the next 1-3 hours, not the actual weather, 
while the other values are valid for (nearly) the current time. BTW: The 
length of the interval depends on the data available for the location, 
so it might be just 1 hour for Stuttgart, but 3 hours for New York and 
even some more for other parts of the world.

When it happens that the end of the interval is chosen for temperature, 
cloudiness etc., then it is because the start of the interval has 
already passed and the download happened after that time, so that data 
at the start of the interval is no more available and only data at the 
end of the interval remains to be chosen.

All in all, these concepts impose certain limits on the intention of the 
plugin. I'm not sure what is the best way to handle this and will 
certainly look into it in the future, maybe interpolating values could 
be a viable way. For now, I believe it will work good enough for most 
people, and the forecasts are updated regularly and data seems to be 
very reliable. If anyone has ideas to improve the current state, please 
let me know.

If that sounds confusing, well....

Harald

-- 
`Experience is the best teacher.'