On Wed, Aug 13, 2025 at 02:57:02AM +0200, Timo Rothenpieler wrote: > On 8/13/2025 2:43 AM, Michael Niedermayer wrote: > > Hi > > > > On Tue, Aug 12, 2025 at 11:22:44PM +0200, Jean-Baptiste Kempf wrote: > > > On Tue, 12 Aug 2025, at 17:59, Kacper Michajlow wrote: > > > > Because of course there is homer-bot... I think everyone seen > > > > marge-bot too https://github.com/smarkets/marge-bot (also GitLab > > > > specific) > > > > > > Be careful, those are 2 very different bots. > > > > > > Marge bot, is a bot to avoid the fight for all the MR to be rebased before merging them. > > > You assign to Marge and it handles the merging at the best time, to save time and cycles to everyone. > > > > > > Homer bot, is a bot to explain clearly the workflow to help what should be mergeable, to avoid exactly the issue we just saw, or the issues we saw in the past on the ML, where people wanted to merge their code quicker than some reviewers were. > > > > > > Homer is configurable, but the idea is basically: > > > - if you are dev/maintainer, and noone comments for x days, you can merge > > > - if you are external/no commit access, you need at least one OK and y days, and one can merge > > > > > - any discussion/thread opened blocks any merging, which allows someone to open a thread to say "I'm reviewing", "I have a doubt" > > > > Theres a cancel automerge button in forgejo > > The automerging in Forgejo is not even close to what those bots do. > It's simply an automatism for "Once all checks complete, and it still > applies cleanly, merge it". > > It does no CI checks again on the final state or anything, which a merge-bot > implementing merge trains/a merge queue would do. You misunderstand what i meant I meant that sending "any reply" causes merging to be blocked is not great design because not every person writing a reply wants to block the merge thx [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB Republics decline into democracies and democracies degenerate into despotisms. -- Aristotle