@corbet @axboe @ljs @brauner

Right or wrong, the feeling I got was that the bar for actions was just raised higher. Now it's not just the CoC committee, you also need a near unanimous vote from the TAB.

https://lore.kernel.org/workflows/87h647yf8r.fsf@intel.com/

@jani @corbet @axboe @brauner cool, more delays in doing absolutely nothing then!

@jani @corbet @ljs @brauner I'm a bit disillusioned at the whole CoC thing - in practice, there are zero repercussions from being a jerk. Banning people from being archived/listed by lore does not prevent them from doing their work, people equating those two things are kidding themselves. Developers getting CC'ed directly generally have no idea the person is banned, and any reply will presumably be archived/listed anyway.

@axboe @jani @corbet @brauner well you might get asked to apologise (no consequences if you refuse)? And get implicit permission to turn up to LF events you're not invited to and abuse people?

Then wait a month to treat the RC process like the merge window again or something?

That's sort of a repercussion right? 🤣

It's all a joke. Kent was the test. Failed.

@ljs @jani @corbet @brauner It does reek of virtue signaling, unfortunately. That's probably not the intent, but in practice it's unfortunately useless.

@axboe @jani @corbet @brauner worse than useless, encourages people to be an asshole imo.

@ljs @jani @corbet @brauner I don't know if it encourages people, but I'm sure some people are tempted to test boundaries (like toddlers...) exactly because there are no practical concerns with doing so. Outside of looking like an idiot online.

@axboe @jani @corbet @brauner Yeah it's only _certain_ kinds of people who will be encouraged.

What infuriated me most about that whole thing was that Kent got to dictate the narrative.

Reality was his broken mm patch was nacked, he threw a temper tantrum and abused the kernel process to take the patch through his tree (outrageous, banning offence stuff imo), and then when it was fixed he got abusive.

His 5,000 word bullshit on patreon + on-list warbling is what people took as the narrative.

@ljs @jani @corbet @brauner Well, as far as I can tell, he might be doing that exact same thing again. His -next branch:

https://evilpiepirate.org/git/bcachefs.git/log/?h=for-next

has a block patch that he sent out to linux-block, got effectively nak'ed as being broken by multiple people, ignored, and then sent out as part of a larger bcachefs patchset and is now staged in the above branch?

@axboe @jani @corbet @brauner oh for fuck's sake.

@axboe @jani @corbet @brauner but hey at least he cc'd you right!

Linus is dropping the ball on this stuff, it's mental he was ok with the mm thing.

I believe he's done this in core fs code before too actually, or snuck in changes to core.

Round of applause to CoC for enabling this and now the rest of us have to put up with it.

Me + colleagues actually often check his PR's in merge window to see if he put any mm stuff there.

So ridiculous to be put in that position.

@axboe @ljs @jani @corbet @brauner if it's breaking -next guess you'll have to ask Stephen to drop bcachefs tree. Also watch out for PR to Linus and reply immediately pointing out there's a nacked/broken commit, and hope for the best. Ideal eh?

@vbabka @ljs @jani @corbet @brauner I replied on the thread, and Kent replied as well. We shall see what happens.

https://lore.kernel.org/linux-block/510692e2-b83c-45bc-8d9d-08f7a172ffe6@kernel.dk/T/#u

@axboe @vbabka @jani @corbet @brauner incredible reply from him 'I know better'.

Same thing as with mm... I think he genuinely believes he can override people

@vbabka @ljs @jani @corbet @brauner that went about as expected

@axboe @vbabka @jani @corbet @brauner ugh man I feel for you having to deal with that shit.

He's so totally dishonest and patronising, it really drives me crazy.

How can you possibly reframe sneaking core changes in via your tree as ok?

It's really insane, I guess ping Linus when he sends the PR.

@ljs @vbabka @jani @corbet @brauner I already CC'ed Linus in that thread so he's aware of it. I'm done replying, nothing good will come of prolonging it.

@axboe @vbabka @jani @corbet @brauner yeah agreed, I don't think engaging with somebody in bad faith is worthwhile.

You've made your position clear, that's the end of it.

@ljs @axboe @jani @corbet @brauner let's hope Linus will do the right thing then :/

@ljs @axboe @jani @brauner So you all know how to contact members of both the TAB and the CoC committee. Everybody involved is kind of trying to figure this out as it happens; constructive suggestions for how things could be improved would be most welcome. What do you think should happen?

@ljs @jani @axboe @brauner Again, suggestions for what $SOMETHING should be would be helpful.

Meanwhile, the most recent CoC action got through the TAB in [checking] two hours and 58 minutes. How much faster do you think a process like this should go?

@corbet @axboe @jani @brauner

Clearly defined escalating responses for repeat offenders responded to a timely manner with clear scope of case ('case still open' creates uncertainty).

Abuse of LF staff not being tolerated.

Refusal to actually do what is asked of you not tolerated.

Also - not treating only 'bad language' as a problem, but also bad behaviour using polite language.

Kent situation is very obviously egregious, it is clearly emblematic of a process fail.

Disagree re 'welcome'.

@corbet @axboe @jani @brauner Went over the $SOMETHING on other thread, (and previously - this isn't empty whining).

I am glad it was streamlined from the ~weeks it took to do something about Kent (and case not closed), but can be forgiven for feeling adding more barriers might result in further delay.

If it's on order of hours, fine. But that's something new right?

@ljs @axboe @jani @brauner *All* of this is something new. Everybody involved is still trying to figure this out.

As far as the other thread: you gave some characteristics that you think the process should have, which is fine. But I have not seen a lot of specifics on how such a response could be implemented within our community. I'll say again: suggestions welcome. Nobody feels that they have a real handle on all of this.

(Just to be clear: I'm not involved in the implementation, but I can do my best to ensure that useful ideas are considered.)

@corbet @axboe @jani @brauner There's a character limit on this instance which limits specificity :)

2 simple ones though:

1. Abuse LF staff, get banned from LF events.

2. Refuse to take action requested of you, punishment either extended or escalated.

Surely the community would accept this... Kent refusing to apologise publicly as was asked of him and suffering no consequences for it was to me, very egregious.

@ljs @axboe @jani @brauner I will pass those on. Of course, attendance at LF events is, in the end, up to the LF. Extending the *kernel* CoC to those events is, by Documentation/process/code-of-conduct-interpretation.rst, explicitly out of scope. That could be reconsidered, naturally.

@corbet @axboe @jani @brauner Thanks.

Has the policy changed there? As I know of at least one case in the past where somebody was told if they did not apologise they would not be permitted to attend an LF event, unless I misunderstood the situation.

@ljs @axboe @jani @brauner That policy hasn't changed since the CoC was adopted.

@ljs @jani @corbet @brauner “We had a situation where the process clearly broke down in -block [...]”

I’m just going to fully ignore this person going forward. They are either deliberately fabricating their own truth, or they truly have no idea how the process works yet plows forward as it must be everybody else that’s wrong. Unbelievable.

@axboe @jani @corbet @brauner can't blame you at all.

He did the same with the mm stuff more-or-less, reframed it as a debate about a GFP flag rather than him intentionally committing what to me is a 'get kicked out of the kernel forever' offence of intentionally working around the process.

The guy needs to go from the kernel, full stop, end of discussion.

I gather there was some consensus reached at lsf that no patches will be taken from him outside of bcachefs?

Which is you know, sane

@ljs @jani @corbet @brauner end of the day, only Linus can make that call as he’s ultimately the one pulling the changes. But in discussion there it seems clear that multiple trees have had this conversation with Linus.

@axboe @jani @corbet @brauner sigh yeah, and then you see how Linus talks to @rostedt - why is he talking to a massively good force in the kernel that way, but tolerating a toxic black hole?

Really it's a Linus problem at the end of the day.

And at some point, Kent is going to do something so beyond the pale that even Linus will have to accept his booting.

Question is how long we let him carry on doing damage until then...

All for a shitty broken file system?...

@ljs

All for a shitty broken file system?...

The FS itself is innocent.

@axboe @jani @corbet @brauner @rostedt

@oleksandr @axboe @jani @corbet @brauner @rostedt well, I base this comment on the beautiful contrary claims of:

1. It is the best thing since sliced bread, more stable than btrfs, better testing than sqlite and the rest of the kernel put together.

2. (after fundamental things break and _development_ is done during rc's) it's an experimental filesystem what do you expect??

I am sure the idea is decent enough, an arrogant implementer ruins all.

@ljs Notwithstanding the personality, it just proves one doesn't have to believe everything that is written on a fence.

@axboe @jani @corbet @brauner @rostedt

@axboe @ljs @jani @corbet @brauner @rostedt hm yes if Linus does nothing/until he does, we can at least try collectively not to reply to him

@vbabka @axboe @jani @corbet @brauner @rostedt sadly he maintains alloc profiling :/

When we spoke about how we need more specific mm M's and R's in the process session you know in this one case, I think we could all agree there are limits lol...