[part 2]
But in addition to the changes of the SLFO platform base, as a “.0” version this was one of the few chances to implement other radical changes; we couldn’t do most of those during the lifetime of a major SLES release. What is released with SLE-15 will need to remain for all the service packs (SLE-15 SP1, -SP2, …, -SP7) until they run out of maintenance (regular as well as LTS).
For example, exchanging YaST2 for something new could only happen now with the new SLES-16.0, or in 10 or 12 years with a SLES-17.0. And the Leap versions run in parallel with those.
Replacing YaST2 has been overdue for a long time. The first version was released with SuSE Linux 6.4 back in late 1999 / early 2000. That’s 26 years ago; aeons in the software world. We have been refactoring parts all the time, but there is a limit what you can do with mere refactoring.
Technologies have changed, people have changed; or rather exchanged. I am the last one in the team who was part of that initial YaST2 release back in 1999/2000.
Many tasks that were once needed to be done in YaST have now moved to the desktops (printer, language, localization, desktop resolution), others are now redundant because they auto-configure themselves (X11, Wayland, sound), others are completely obsolete.
What remains is primarily the installer, and people have been clamoring for something web-based since forever. The young kids love web programming, and they learned it in unversity. So a web installer it was, initially started as a SUSE Hack Week project. And Agama was born.
Yes, many aspects are new. It’s evolving. The team as a whole has the know-how about the Linux installer in general, what works and what doesn’t, what old stuff to throw overboard.
The users have to learn those new concepts. Sometimes prior knowledge of how YaST worked is a real obstacle; we found that often enough new users have no problems because they don’t need to unlearn things that YaST did differently.
Some parts like the storage setup has already been refactored after users were too often confused. Some other parts will probably still need refactoring. It’s all work in progress.
So, if there are major frustrations, please don’t stop speaking up, so the project can keep evolving.
System administration after the base installation is a major pain point. Cockpit can do some things; for other things it can be limiting.
Case in point: Package management. That’s why I chose to rip out that part of YaST to create Myrlyn; it has always been kind of a separate part for technical reasons (and I had written it in two iterations before anyway, before libzypp and once again when libzypp arrived).
Myrlyn is libyui-qt-pkg a.k.a. YQPkg plus some init code to load the repos, and the transaction commit phase where packages are actually downloaded and installed. And some dozen bug fixes because YQPkg had been kind of neglected for many years.
Other pain points include the bootloader configuration; our Cockpit team added that just recently. Maybe they’ll add some more things.
OK; enough for now. 