Linus Torvalds Begins Expressing Regrets Merging Bcachefs

[phoronix]

Phoronix: Linus Torvalds Begins Expressing Regrets Merging Bcachefs

There's been some Friday night kernel drama on the Linux kernel mailing list... Linus Torvalds has expressed regrets for merging the Bcachefs file-system and an ensuing back-and-forth between the file-system maintainer...

https://www.phoronix.com/news/Linus-Torvalds-Bcachefs-Regrets

[Quackdoc]

Bcachefs has been pretty chonky patches, I wonder if it wouldn't be better to just distribute DKMS modules if they need so much code to get in during the "fix" cycle.

[geerge]

Kent needs to stop trying to upstream so quickly. Follow the rules and don't piss off the maintainers above you, easy?

[fong38]

Bcachefs has been pretty chonky patches, I wonder if it wouldn't be better to just distribute DKMS modules if they need so much code to get in during the "fix" cycle.

Not a bad idea. Provide a dkms module for bleeding edge, upstream commit state to kernel (at appropriate development cycle time points) once deemed stable enough. Not sure how much more burden it would place on him though.

[nadir]

Having read through the thread, it really seems like Kent still just doesn't get it.

He seems to have serious I-am-the-main-character-syndrome. People point out why process is important and that his actions mess up things for the other people he works with and his response boils down to "yes, but my code is better and much more important, so the rules don't apply to me".

I wasn't following bcachefs development this year, but it seems like Kent has not gone through a fundamental change in attitude since the drama last year.

[spicfoo]

Pretending that there is consensus that Bcachefs in it's experimental state is more robust than Btrfs and even XFS is a very bold claim and is mixed messaging. Even ZFS long after it was in production usage for years ended up with multiple data corruptions bugs. It's one thing for fanboys to do this but when developers say this, it is a much bigger concern. Without robust usage for varied uses by Linux distributions, it is impossible to say how robust it is. If you are playing with it, have multiple backups and do not trust that it is not going to have data loss bugs in there.

[blackiwid]

I don't get the threatening language, why escalate it, Linus is the dictator he can just say no I don't take this patch, DONE.

Why does he have to add a threat to kick him out, or if he regrets his decision which he did, so only he is guilty of that decision. Say that to your mirror or think it, but don't say it.

This is very abusive behavior, when I think a boss would publicly shame one of their main employers it would be unacceptable especially over practically nothing, just because he wants to reject a suggestion of the employer.

Maybe this article omits some forth and back where Kent became ugly or threatened themself, but I see here bad leadership, imho.

What is his problem does he not want to say no? Or not make the decision he himself believes that everybody knows this is experimental so just don't accept it and keep it in a buggy data eating state... either you don't believe what you yourself say and people expect it to be somewhat stable or the decision is easy.

[npwx]

This guy will never learn, I said it before and I'll say it again, btrfs will win simply because it's not causing problems with the development process and it's not like development has stalled.

Indeed. Once again he wants to be special and does not even stop after explicitly being told so.

Bcachefs, the filesystem than won't eat *my* data.

[Topolino]

Happens all the time. Developers need to follow the rules and expectations set by Upstream.

[NobodyXu]

Maybe it's time to consider NILFS, while it's not for RAID/data center/NAS, it seems pretty awesome for desktop usage with automatic snapshot per save/flush.

And the automatic snapshot can also be pinned to prevent it from being automatically removed..