On Wed, 2025-05-28 at 09:26 +0530, Aditya Kumar Singh wrote: > > Concur with your thoughts here. However, I just want to bring one point > related to _Dependencies_. Yes, nipa doesn't handle anything wrt. dependencies. It's also tricky - consider if it has recursive dependencies, does it potentially go into a loop to apply patches? Where does it take them from? I'd think only the same patchwork instance, but it feels subject to feature creep. > In case of Dependency (Depends-on tag) between series (even from the > same tree), the bots - kernel as well as the NIPA, currently does not > handle it gracefully and in such cases there could be legitimate build > failures reported since obviously the declarations are in parent series > which is not taken by the bots. So in such cases, _red_ items will be there. Right. > Do you have any suggestions on how we should go about handling > dependencies? So Konstantin also said the "Depends-on:" tag likely wouldn't fly, and I agree, why should this be recorded in the final git commit message. Looks like b4 supports something else though: https://lore.kernel.org/all/20250528-sparkling-expert-taipan-5ad429@meerkat/ Might be worth standardising on. I did plan to move to b4 eventually, but mostly when I actually do merging I'm in a hurry and/or more focused on reviewing, so haven't focused on the workflow much yet. With Miri now sending pull requests for iwlwifi things got simpler though. As to whether or not the bot will get support for it - I don't know, I guess we could take it up with netdev. Maybe you can float it as an idea for the netdev foundation [1] if development work is there - but maybe also first search netdev to see if it's been discussed before :) [1] https://github.com/linux-netdev/foundation For now I guess I think reposting the patchset after the prerequisites land isn't _horrible_? johannes
