On Sun, Nov 22, 2020 at 08:29:22PM +0100, Christian Eggers wrote: > this was also not by intention. Vladimir found some files I missed in the > first series. As the whole first series had already been reviewed at that > time, > I wasn't sure whether I am allowed to add further patches to it. Additionally > I didn't concern that although my local build is successful, I should wait > until the first series is applied...
When I said that, what I was thinking of (although it might have not been clear) was that if you send further patches, you send them _after_ the initial series is merged. Alternatively, it would have been just as valid to resend the entire series, as a 3+3 patchset with a new revision (v3 -> v4), with review tags collected from the first three, and the last 3 being completely new. Jakub could have superseded v3 and applied v4. The idea behind splicing N patches into a series is that they are logically connected to one another. For example, a patch doesn't build without another. You _could_ split logically-connected patches into separate series and post them independently for review, as long as they are build-time independent. If they aren't, well, what happens is exactly what happened: various test robots will report build failures, which from a maintainer's point of view inspires less confidence to apply a patch as-is. I would not be surprised if Jakub asked you to resend with no change at all, just to ensure that the patch series passes build tests before merging it.
