Fix i18n regression

[Parker Coates]

On Mon, Sep 13, 2010 at 18:30, Albert Astals Cid wrote:
> A Dilluns, 13 de setembre de 2010, David Faure va escriure:
>> On Monday 13 September 2010, Albert Astals Cid wrote:
>> > klocale over a fake component being created for
>> > some reason before the klocale with the correct component
>>
>> I am very curious about the "for some reason" :-)
>>
>> What's calling KGlobal::locale() before a main component is set?
>
> In case of kpat is an invalid i18n call before the KApplication is created, in
> other cases it might be the same. Of course those cases should be detected and
> fixed but the punishment you get for the error is too big if you ask me.

I just fixed the issue in KPat. [1]

I also agree that this is a great use case for a big, ugly warning.
Silently breaking all translations isn't cool. :)

Parker

[1] http://websvn.kde.org/?revision=1175050&view=revision