Comment # 5 on bug 1163737 from Jonathan Kang

(In reply to Nathaniel Graham from comment #4)
> It's unfortunate that the provider and user of the API have such different
> expectations.
> 
> For what it's worth, I mildly lean toward Richard Hughes' position from a UX
> perspective; users of a GUI updater app will, 9 times out of 10, not know
> what do to do resolve packaging conflicts requiring interactivity anyway. I
> think it would indeed be best to either do the most obviously correct thing,
> or at least the most non-dangerous thing. The point that requiring
> inactivity breaks unattended updates is also a valid one.
> 
> Relatedly: in my experience on Tumbleweed, cases where interactivity was
> required during an update were caused by instances of broken packaging, most
> commonly de-sync between the Packman repo and the main repo. I've since
> moved to Fedora KDE with the RPMFusion repo (which serves a similar function
> to the Packman repo in openSUSE land), and I have not once experienced the
> kind of repo de-sync problems that required manual interactivity there, or
> in fact any kind of broken packaging issues at all, ever. So it also seems
> possible to minimize the condition causing this requirement in the first
> place, at least. And from a user perspective, that's the best solution.

I do agree. There is definitely room to improve at openSUSE on how to properly
distribute software/packages in various repositories, so that such issues can
be minimized.