On czw, Apr 11, 2019 at 11:31 PM, "Maurizio Galli (MauG)" email@example.com wrote:
On Wed, 10 Apr 2019 at 11:09 PM, Ludwig Nussel firstname.lastname@example.org wrote:
Maurizio Galli (MauG) schrieb:
On 4/9/19 7:17 PM, Ludwig Nussel wrote: [...]
I think we reached a fair compromise with the primary choices
in the role selection while all other options are available via
"custom" option. I'd prefer to see the patterns cleaned up so
presentation in YaST can be done in a meaningful way¹ rather
abusing that role selection dialog for even more stuff.
Just to give you a bit of background, I was personally not particularly enthusiastic of the "generic desktop" option because
adds additional steps and is not as intuitive as the older "custom install", which I liked a lot. A few others in the online
I reside expressed similar concerns which eventually was one of the reasons that motivated me to do the work and push for a Xfce system role, which has been done for TW.
Anyway I appreciate your time in explaining things to me as there
obviously many things I don't know yet and I'm learning along the
I will probably bug you again about this when the new Leap 15.x or
What's basically needed is development resources from the yast team and a UX person. We need a solution that works for SLE, Leap and Tumbleweed without too much different code. The previous desktop selection was too openSUSE centric and the role selection coming from SLE on the other hand isn't really geared towards openSUSE.
Sounds like what we basically need is LCP who knows UX and is familiar with Yast :-)
I'm sure there are more UX experts than just me out there, I only know YaST as much as I looked into it, which is too much :P
I have seen some interesting mockups that if technically doable would be worth exploring.
Doable with a lot of work in various places, considering YaST is split up into a lot of pieces.
Desktop environment selection replaced with roles was a blessing in disguise in my opinion, there is an opportunity there to remove the useless pattern dependencies, patterns depending on or recommending other patterns, which is an issue primarily on Tumbleweed where recommends get reinstalled with dup (the same happens with Leap, but Leap dups are rarer). The whole recommends section of most patterns should be moved to be role-installed patterns instead. This way a person that selects a gnome pattern, but does not want games that are recommended by it, can uninstall it easily without worrying about them coming back on the next update. The obvious disadvantage is that the way roles work prevents us from having a software selection, because a role has to be selected first for any pattern recommendation to be applied. Another obvious issue is that removing a role doesn't remove the software that came with it, but that's an issue to be resolved with zypp itself (most likely via introducing similar database as is present in dnf, to log all the package activities and dependencies so we can have proper removal of needless dependencies).
But I am not sure if we should keep the discussion here or bring it to one of the ML.
I mean, the people that should know about this are YaST folks primarily. They have been working on a bunch of cool stuff recently, which enhances quality of YaST's various backends. The work that remains to be done is the frontends. libyui ncurses, qt and gtk need love since the day they were created, it would also make manatools developers happy to know somebody cares to improve on the toolkit abstraction that they also use. GTK backend especially needs some special care and attention, since it could be a great thing to ship with GNOME/Xfce when it's usable (and there are people like Besser that put their heart and soul into that backend over the years to still keep recieve chronic lack of attention from YaST guys).
The mockup in question: https://cdn.discordapp.com/attachments/496049131928682506/564392720513105947... (Also mentions the projection of when this will get implemented based on Leap version)
LCP [Stasiek] https://lcp.world