On Thu, Feb 15, 2024 at 3:18 AM Mimi Zohar zohar@linux.ibm.com wrote:
On Wed, 2024-02-14 at 16:21 -0500, Paul Moore wrote:
I'm not a big fan of sharing topic branches across different subsystem trees, I'd much rather just agree that one tree or another takes the patchset and the others plan accordingly.
Just curious why not?
I don't like the idea of cross-tree dependencies, I realize the term "dependency" isn't a great fit for a shared topic branch - no one needs to feel the need to explain how pulls and merges work - but it's the conceptual idea of there being a dependency across different trees that bothers me. I also tend to dislike the idea that a new feature *absolutely* *must* *be* *in* *a* *certain* *release* to the point that we need to subvert our normal processes to make it happen.
Further, I believe that shared topic branches also discourages cooperation and collaboration. With a topic branch, anyone who wants to build on top of it simply merges the topic branch and off they go; without a shared topic branch there needs to be a discussion about which other patches are affected, which trees are involved, who is going to carry the patches, when are they going up to Linus, etc. As someone who feels strongly that we need more collaboration across kernel subsystems, I'm always going to pick the option that involves developers talking with other developers outside their immediate subsystem.
Hopefully that makes sense.
Based on our previous discussions I was under the impression that you wanted me to merge this patchset into lsm/dev, but it looks like that is no longer the case - which is okay by me.
Paul, I don't recall saying that. Please go ahead and upstream it. Roberto can add my acks accordingly.
I believe it was during an off-list chat when we were discussing an earlier revision of the patchset, however, as I said earlier I'm not bothered by who merges the patches, as long as they eventually end up in Linus' tree I'm happy :) I *really* want to stress that last bit, if you and Roberto have stuff queued for the IMA/EVM tree that depends on this patchset, please go ahead and merge it; you've got my ACKs on the patches that need them, and I believe I've reviewed most of the other patches that don't require my ACK. While there are a some LSM related patches that would sit on top of this patchset, there is nothing that is so critical that it must go in now.
If I don't hear anything back from you, I'll go ahead and merge these into lsm/dev later tonight (probably in about ~12 hours from this email as I have some personal commitments early this evening) just so we can get them into linux-next as soon as possible.