Staying on GTK/GNOME 3.8 next cycle/for the LTS?

[Sebastien Bacher]

Le 02/10/2013 01:45, Tim a écrit :


>
> Either way some concerns I have
> - Ubuntu will keep holding back on GNOME updates until QML/Touch stack is ready and then just dump it
That's not going to be the case, I think that staying one release behind 
is a fair tradeoff and that we should keep doing that. One reason for 
the stronger pushback on the update is that, in my opinion, we should 
hold back on controversial changes for the LTS. If we get some hit, from 
unhappy users, after the LTS that's fine ... they can stay on the LTS or 
we have time address those during the next LTS cycle.

> Transitioning from 3.8 -> 3.12 would likely be a big nightmare, also if we end up with a 2 cycle divergance, mixing packages from
> different releases will become much harder than it already is.

The transition shouldn't be harder, it's basically:
- update the libs (they are api/abi compatible so it's fine)
- update the apps
- update the components like g-s-d/g-c-c in sync

That shouldn't be more complex that usually, especially if the patch 
rebasing work happening in a ppa during that cycle
>
> - We will most likely need to transition ubuntu GNOME to wayland at some point, however we can't really even start on that in an experimental
> capacity until 3.10 is in the archives.
Even if Ubuntu was going to go for wayland (which is not likely the 
case, at least for Unity), that's enough changes that it wouldn't happen 
before the coming LTS. It makes sense to start those sort of transition 
at the beginning of a LTS cycle...
>
> - There are a number of major bugs we have on the PPA's that are really outside of our scope to fix, but as long as they are PPA only packages,
> no one cares to help fixing them. Things like the Software Center crash with updated Webkit plus a few new issues introduced with 3.10 such as
> unity custom menus in GTK and the DisplayConfig needing to be implemented in Unity.
I don't think that's true, and that's another reason for not wanting to 
go with the update. Those bugs don't get ignored because they are in the 
ppa, they are not addressed because nobody has spare cycle to work on 
those. Landing the update in the archive would lead to a situation where 
you would increase the stress level on people who are overworked 
already, and wouldn't get half the bugs looked at anyway in return. It's 
a no-win situation for everyone...
>
> - Its really unlikely that we will be able to track 3.12 on a 3.8 base, we mostly get away with 3.10 since some of the core libraries in Saucy
> did get updated to 3.10 versions, however there are packages we simply can't package on the PPA's such as glib, gvfs, cogl/clutter etc due the
> massive list of rdepends. Right now we have had to revert a huge number of patches just to get gnome-shell 3.10 running on Saucy.
Right, that's one cycle only though, I don't think it would be the end 
of the world to hold back for another cycle, knowing that the net 
benefit is better stability for our users
>
> - If the PPA's end up a cycle behind and there is complete lack of wayland support, we will likely start loosing users to Fedora etc.
To be honest I fail to see "running on wayland" as a something users 
want. Especially that the current goal is to reach parity. If things go 
perfectly, the next GNOME version is going to run as well on wayland 
that it is on xorg. It's going to be a win for the future, but probably 
not something that makes any day to day difference to users. By the time 
"GNOME on wayland" is ready, the LTS is going to be out. Sure, it might 
be an issue for some tech users who want to be on top of the most recent 
changes, but I don't think they are the primary target of a LTS version...


Cheers,
Sebastien Bacher