Sorry if I ask some stupid question, as my experience with Tumbleweed is limited: I tried it several times with previous openSUSE releases, but always stopped using it sooner or later upon facing some incompatible update (such as when libc or another fundamental library is changed, causing a zillion package version conflicts). Anyway, I have accustomed to Tumbleweed always being far ahead of the static branch: for example, even if openSUSE N.x + TW is kept unupdated for a month or so before openSUSE N+1 release, there still was a need to wait some time more to catch up and so that most packages do not get downgraded.
However, after installing openSUSE Leap 42.1 from scratch and updating it, then trying to migrate to Tumbleweed repositories, I found out that would require to downgrade almost 600 packages. One such notable example is Firefox: while Leap 42.1 offers Firefox 45.0 for some time already, Tumbleweed is still with Firefox 44.0. I don’t get is: isn’t Tumbleweed supposed to provide the most current and stable experience?
Well, Tumbleweed as a true rolling release is expected to be generally more up-to-date, but not “stable” by any meaning, since it receives major upgrades every week or so…
That said, my understanding is that TW has always a newer kernel and generally newer system packages (e.g. systemd, system libraries…) and newer desktop environments (e.g. Gnome 3.19 vs. Gnome 3.16 for Leap).
That newer environment might involve more testing / debugging / tweaking to some applications to make them reliable enough for productive work.
So it is not unusual that a few user applications like Firefox take longer to appear in TW if problems arise.
On the other hand, e.g. LibreOffice is stuck to 5.0.x and is receiving only bugfix releases on Leap while the 5.1.x series already landed on TW.
Members more familiar with the TW development process might add to this, I’m only a casual user of Tumbleweed so far.
Has the twiki been updated to reflect this? I thought I understood was TW was, but recently I am confused about exactly how much Q/A actually happens before push to repos. I feel like minimal or zero testing for major packages / kernel / platforms.