ALP-Leap-replacement Architecture Meeting - https://en.opensuse.org/openSUSE:ALP/ArchitectureTeam Meetings take place on https://meet.opensuse.org/meeting and notes are available at https://etherpad.opensuse.org/p/weeklymeeting Participants: Richard, Lubos, Valentin, Marcus, Bernhard, Phoeinx, Knurpht, Rob, Emiliano, Ish, Dirk 1. Update from Richard: Communications - There has been NO communication on https://matrix.to/#/#alp:opensuse.org since July 1 Q from Richard - have discussions been happening elsewhere? A- There hasn't been much, but there has been some conversations. Richard needs to fix his Matrix client ;) but more people should join and discuss also. Ongoing efforts - there are currently 2 concepts for an ALP-Based-Leap Replacement https://en.opensuse.org/openSUSE:ALP/ArchitectureTeam#Experimental/Research_... Linarite update - No progress yet Slowroll update- "I did a snapshot of a random Tumbleweed version (20230705) and setup OBS projects to build a package update ontop of it. That worked with some help from Adrian. During that we discovered that we will probably need an own project like openSUSE:Slowroll to have our own prjconf that can move independently of Factory/Tumbleweed. https://build.opensuse.org/projects/openSUSE:ALP:Experimental:Slowroll/meta I also started https://github.com/bmwiedemann/slowroll-tools that already provides data for a core-ness value for a given package. More tools should be added there later, e.g. to do heuristics on .changes files and code diffs to provide input to the required risk+benefit-assessment. One open question is if we want multiple overlapping slowroll versions. If most work is automated, that should not be much extra effort. Then we could have a slowroll-2023-07 followed by a slowroll-2024-01 with overlapping support periods so that sysadmins can dup on their own schedule. It could be like openSUSE before Leap (but with more automation to keep the work easy). I would like that." 2. Discussing concepts for additional Experiments/Research Projects No additional concepts raised 3. Any other buisness - Group mail from Max, is this another concept? Richard - doesn't seem to be, but does provide many good questions the group needs to consider. - Was recurring suggestions via YouTube and Reddit that openSUSE may not need a Leap replacement. What do we think as a group? A- To be kept in mind and maybe used as a question in poll (see below) Lubos - would like to see a solution for community to be able to touch the maintenance workflow Marcus - the group needs to define what the Leap replacement looks like before worrying about how to maintain it Richard - exactly - well if we did Slowroll, then we wouldn't have a maintenance workflow like the current Leap at all Phoenix - What does lifecylce mean in the context of Slowroll? Richard - Something along the lines of how often/how quickly after a package is in Tumbleweed does it appear in Slowroll Phoenix - But the metric of how often is a package updated is mostly irrelevant for a point release. Users expect a system to be installed and then mostly stay stable. The important question is "are packages maintained" and that does not necessarily align with packages being kept at bleeding edge Dirk - Also Packages in SLES are maintained which means, they will be updated when there is a reason for them to be updated. Lubos - Max tries to update all community packages in Leap 15.X few times during the development cycle. We have it as part of our release process. It's true that some of the SLES packages were not updated since SLES 15(.0):GA, these receive on demand updates typically requested via Jira or https://code.opensuse.org/leap/features Dirk - suggestion to create a poll to gain feedback on which concepts folk might like https://en.opensuse.org/Portal:15.5/Retrospective has plenty feedback on Leap 15.5 pros+cons Ish - the poll should be really specific, and it should be done for Members only, using the election platform. Ish volunteers to run it. AI - Richard, Lubos, Dirk to work on wording for Ish. Poll is likely to focus on the concepts we have on the table, namely Slowroll, Linarite, and not-replacing-Leap. Phoenix - Shares some thoughts - Leap users like Leap because they like a point release. If we give them a rolling release they'll probably just go to a different distro. Contributing to Leap is very difficult and gets in the way of a point release. The main thing contributions want to do is just add new packages, so, why not have a concept which is an expanded packagehub, which a much simpler contribution process. Richard - cool idea..it's not really different from openSUSE:Backports and would likely face the same challenges gathering contributions Lubos - you'll hit the problems we already hit..new package requires an update to FOO which blocks the change until the dependency can be updated in the next point release Richard - Exactly, this is why Tumbleweed has the concept of "To be able to change any one thing, we need to be able to change everything". Any point release model will effectively block such contributions for months or years at a time. Next meeting - after pool closing, TBA -- Richard Brown Distributions Architect SUSE Software Solutions Germany GmbH, Frankenstraße 146, D-90461 Nuremberg, Germany (HRB 36809, AG Nürnberg) Managing Directors/Geschäftsführer: Ivo Totev, Andrew Myers, Andrew McDonald, Martje Boudien Moerman