On 27/03/2025 17:44, Wolfram Sang wrote: > >> You did not object to last discussion about this (a month ago) - neither >> to my comments nor to resolution - so this patchset repeating the same > > Because I cannot follow every Renesas patch series there is. You are > long enough around to know that large companies have different entities, > groups whatsoever. It is quite a challenge to streamline this via one > group, we need to share work. We do try hard, though, and have a > ARM/RISC-V/RENESAS ARCHITECTURE maintainer. Geert does a *hell of a job* > getting all these submission into shape, and he surely does not accept > code thrown over the wall. And geez, the patch series was just sent > yesterday, you didn't give us even time to raise the issue internally. > >> pattern from the same folks while ignoring previous talk is >> contradicting "not too bad at fixing stuff". > > First, being a maintainer myself, I do understand the frustration of > patch review not being honored. I can also agree that this series did > not work out perfectly. But that does not mean that we don't care, in > general. Despite all imperfection and possibly different opinions, we > try hard to be a good citizen and spend considerable time on doing > things right. Accusing us of throwing just "code over the wall" because > there is an issue somewhere which hasn't been worked on in one month is > plain unfair. We do not speak about same things. I speak of review being ignored for multiple revisions in one patchset and then another patchset sending exactly the same pattern. Previous patchset receive my review about this. Thierry ignored it and send v2 with same code. Then v3 with exactly the same code, but with a remark in cover letter "but such a change is out of scope for this patchset." And now Pabhakar sends the same pattern. Each of these contributors were not changing here anything, it's like not their job. It looks like this will never get fixed, because each person wants to just get their stuff merged, so let's ignore the reviewers comments. That's not how upstreaming works - you need to change some things, fix some stuff, add more code, if you want to add your independent features. That is how upstream was always. The easiest example is - one new driver for some completely new feature is fine. Second new driver for similar new feature receives feedback: please create subsystem to have common set/handling of that new thingies. Best regards, Krzysztof
