> there are parts of a linux distro that tie in with lots of other things in the distro and are rather hard to upgrade independently (things like KDE, GNOME)
They should be able to upgrade independently. Right now I'm using k3b, ktorrent, kid3 and digikam. I've never had problems with any of them.
Posted Sep 7, 2010 20:28 UTC (Tue) by dlang (✭ supporter ✭, #313)
[Link]
I would not consider any of those apps as ones that would be hard to update.
however a new version of QT, GNOME, KDE, glibc, etc or changing what compiler is used to compile the distro would require a LOT of testing to make sure that it works well with everything else. These are the types of things that I think work well on a six-month upgrade cycle where several of these things get udated at once as opposed to a full rolling-update distro like gentoo where these may appear at any time.
Canonical
Posted Sep 7, 2010 22:42 UTC (Tue) by nicooo (guest, #69134)
[Link]
My point is that most of the apps that kde uses could be released individually since there are already some that do so. Also, IMHO it's a lot harder to test when everything is released at the same time. If a program crashes after updating a single library it's easier to find the problem.
Canonical
Posted Sep 7, 2010 22:51 UTC (Tue) by dlang (✭ supporter ✭, #313)
[Link]
I have no problem with kde apps being shipped and updated separatly from KDE (or gnome apps separatly from GNOME)
I think this would be a very good thing.
there is desktop infrastructure, and then there are desktop applications. Most applications can be used on either desktop nowdays, and I personally think that the attitude that you must use KDE to use any KDE apps (or GNOME to use any GNOME apps) is preventing competition between the apps developed for the different desktops.
Canonical
Posted Sep 9, 2010 14:46 UTC (Thu) by dgm (subscriber, #49227)
[Link]
Yes and no. Releasing everything together has is benefits too. You only need to test one version of every app, for instance. Having to test many versions of every app for regressions before updating a library increases the testing burden quite a bit.