Reminder: KComponentData changes (please help test!)

[Matthew Woehlke]

So far there have been no objections to generalizing the ability to 
create KComponentData instances that do not become the global instance 
(as currently happens if name=="kdeinit4") (1). Because there have been 
no objections, I still plan to commit this Monday, but it would be 
really nice if I had more test results. Marijn Kruisselbrink (OSX) 
considerately tested this patch (thanks again!) and found no blatant 
issues, but it would be really great if someone with Solaris could also 
test it (Linux tests can't hurt either but we're more worried about the 
more "exotic" platforms).

Why change KCD? In KDE3, styles could not use kdeui. The crashes seem to 
be fixed in KDE4, so the remaining problem is that no KComponentData 
exists in Qt-only applications unless the style creates one. This patch 
allows that to happen in a way that seems safer than risking the style 
becoming the global KCD instance (at any rate, Matthias Kretz preferred 
it, and I'm not inclined to argue).

If you have time to try the patch, testing directions are at (2). You 
need to apply both patches; (1) changes KComponentData, (2) changes 
Plastik to take advantage of (1). Also, Mek noted that the kinit_win.cpp 
hunk of the patch (1) doesn't seem to want to apply, probably due to DOS 
line endings. Unless you are on Windows, presumably it is safe to ignore 
this failure :-).

[1] http://permalink.gmane.org/gmane.comp.kde.devel.core/43734
[2] http://permalink.gmane.org/gmane.comp.kde.devel.core/43735

-- 
Matthew
"Resistance is futile" -- Borg
"No, resistance is the reciprocal of conductance" -- Dave Korn