Linus Torvalds Reviews The Bcachefs File-System Code

[phoronix]

Phoronix: Linus Torvalds Reviews The Bcachefs File-System Code

The long-in-development Bcachefs file-system driver was submitted for Linux 6.5 but never merged this cycle due to various technical issues and developer in-fighting. Linus Torvalds himself has now gotten around to reviewing the proposed code and chiming in on the situation...

https://www.phoronix.com/news/Linux-Torvalds-Bcachefs-Review

[zcansi]

Glad to hear there is some progress and a not-unfavorable view by Linus. Bcachefs always looked interesting to me but I was expecting some problems when it came time to merge it, as that is pretty much always how it goes when a developer who has been working separately for a long time shows up with a whole pile of new code.

[gnattu]

As Linus said the part out of bcachefs looks okay, the long debate over where to place such code could be settled and we can move on to get real progress.

[Crocodilus]

Finally ol’ Linus got up that long comfy chair and did something. He seems kinda lazy in these past few decades .

[avis]

Glad to hear there is some progress and a not-unfavorable view by Linus. Bcachefs always looked interesting to me but I was expecting some problems when it came time to merge it, as that is pretty much always how it goes when a developer who has been working separately for a long time shows up with a whole pile of new code.

The problem is not his massive drop of code and that it's been developed out of the tree for a long time, the problem is that he introduces some non-trivial invasive changes to other kernel subsystems and people are naturally wary of that. Bcachefs could be great but it doesn't mean it should blow the kernel up in your face if you don't even use this filesystem.

And the second issue people are not even talking, is the bus problem. Kent Overstreet is the sole developer of this massive FS. This doesn't bode well if anything is to happen to him.

[Namelesswonder]

Probably not going to making it into 6.6, given that there will be discussions of further review and quibbles on the parts of the patchset not directly composing the filesystem.

However 6.6 would be the proper test round for those parts, to quell the "merge bcachefs and you'll break the kernel" argument, leading to 6.7 for implementing the full filesystem with plenty of time to test.

[gnattu]

The problem is not his massive drop of code and that it's been developed out of the tree for a long time, the problem is that he introduces some non-trivial invasive changes to other kernel subsystems and people are naturally wary of that. Bcachefs could be great but it doesn't mean it should blow the kernel up in your face if you don't even use this filesystem.

And the second issue people are not even talking, is the bus problem. Kent Overstreet is the sole developer of this massive FS. This doesn't bode well if anything is to happen to him.

This is a bigger issue that I'm more concerned than "where to place the code". It is hard for others to understand this massive implementation and become another maintainer. I don't know if Kent can find a team or already having a team. But Kent also said in the mailing list that big corporate folks(i.e RedHat) cannot give much support to him unless this is getting merged. So it looks like a chicken and egg problem. I know a lot of people hate RedHat for whatever reason, but for such a critical and massive filesystem, we do need support from more than a single talented person in the long term.

[cl333r]

Finally ol’ Linus got up that long comfy chair and did something. He seems kinda lazy in these past few decades .

That happens when you get promoted from coder to decider and over time realize that the project moves fine along without you and you just have to write summaries of releases. I don't see him as the leader or founder anymore, he's more like the lazy press secretary on payroll by big corporations.

[Crocodilus]

That happens when you get promoted from coder to decider and over time realize that the project moves fine along without you and you just have to write summaries of releases. I don't see him as the leader or founder anymore, he's more like the lazy press secretary on payroll by big corporations.

I couldnt have said it better myself . He doesnt even decide,maybe for small trivial things. Or he just rambles to keep an appearance of leadership and authority.

But the big ones,are pushed by big corpos that pay him to pick his nose all day long

[cynic]

This is a bigger issue that I'm more concerned than "where to place the code". It is hard for others to understand this massive implementation and become another maintainer. I don't know if Kent can find a team or already having a team. But Kent also said in the mailing list that big corporate folks(i.e RedHat) cannot give much support to him unless this is getting merged. So it looks like a chicken and egg problem. I know a lot of people hate RedHat for whatever reason, but for such a critical and massive filesystem, we do need support from more than a single talented person in the long term.

probably RH isn't interested in bcachefs the same way they're not interested in btrfs