Dear Tumbleweed users and hackers,
This week felt quite unspectacular, but maybe that’s just for me and because I already know what will expect us next week. And so will you, soon! Anyway, first we look back one week. Tumbleweed has seen 5 published snapshots (0513, 0514, 0515, 0517, and 0519).
The main changes in those snapshots included were:
- Linux kernel 5.12.3
- elfutils 0.184
- sudo 1.9.7
- KDE Gear 21.04.1
- PostgreSQL 13.3
- Switched from go 1.15 to go 1.16
And now, for what we plan on releasing next week:
- NetworkManager-Applet 1.22.0 (snapshot 0520, if all goes well)
- Linux Kernel 5.12.4 (also snapshot 0520, if all goes well)
- Rebuild all of Tumbleweed using GCC 11 (will happen on Tuesday, May 25th)
- UsrMerge: as we already have a full rebuild scheduled, we’ll also flip UsrMerge on. This means the first snapshot published after May 25th will change /bin to be a symlink to /usr/bin. Thanks to Ludwig for having the energy to push this over the finish line! The switch started long ago
[…] #Tumbleweed – Review of the week 2021/20 http://dominique.leuenberger.net/blog/2021/05/opensuse-tumbleweed-review-of-the-week-2021-20/ #suse #gnu […]
I hope next Network Manager goes well and fix a bug related to openVPN (https://bugzilla.opensuse.org/show_bug.cgi?id=1186091)
Good work !
Talking about a “cold week” , I would like to understand why since 2021-04-28, the following releases had a so low rate according bombatower (https://review.tumbleweed.boombatower.com/).
Can you help me to understand what caused a row of “not so good” releases ?
Miguel,
I asked as similar question a week or so ago and was told that the most of the failures were related to kfence being enabled in the Kernel causing many tests to fail because the system did not return to idle state fast enough for the tests.
I believe they reversed that change around the 20210511 build as that’s when the number build tests that passed jumped back up and kfenced being disabled was also mentioned in the Review of the week 2021/19.
Hope that helps!
Joe
This web site is maintained by a community member and tries to be smart about whatever metrics.
From the release manager point of view, it does not have any relevance and does not represent our view on the quality of any snapshot.
Keep in mind that soft fails are not failures in most cases: we mark a softfail only if it is either possible to workaround it or if the issue does not really impact the user. That idle best was in the latter category: the system was totally usable and that high idle counter did not cause any real issues for the user.
Dominique,
Thanks very much for that explanation on softfails.
Two questions:
1) What would you recommend we use to determine the quality of a snapshot?
2) Is there a resource available that would provide a summary and make it easy to see that all those softfails were because of high idle counter or do we have to look at each of the test results to come to that conclusion?
Thanks for doing the “Review of the week” I have been following it once I found your site!
Joe
.