I think this change is a bit too intrusive for an SRU from testing/assessing the potential regressions POV and also does not fully resolve the redundancy between tasks and meta packages, just special-cases two of them.
I think the minimal fix of skipping just systemd-timesyncd either in u-r-u or in the seed generation for focal only would be a more appropriate to limit the impact.
For groovy and later I think the task generation should be changed to not pull in dependencies to tasks and as a result u-r-u could keep handling tasks the way they used to be handled.
I'm OK with picking this change and SRU-ing it because the change is not wronk, but I think it is between a minimal fix and a general solution and this makes it not optimal in my eyes.
I think this change is a bit too intrusive for an SRU from testing/assessing the potential regressions POV and also does not fully resolve the redundancy between tasks and meta packages, just special-cases two of them.
I think the minimal fix of skipping just systemd-timesyncd either in u-r-u or in the seed generation for focal only would be a more appropriate to limit the impact.
For groovy and later I think the task generation should be changed to not pull in dependencies to tasks and as a result u-r-u could keep handling tasks the way they used to be handled.
I'm OK with picking this change and SRU-ing it because the change is not wronk, but I think it is between a minimal fix and a general solution and this makes it not optimal in my eyes.