It’s becoming easy to see why Linus didn’t merge anything from bcachefs for 6.17. And Kent isn’t gaining himself any supporters by tearing down other filesystems in his tantrum.

  • cadekat@pawb.social
    link
    fedilink
    arrow-up
    5
    ·
    17 hours ago

    Huh, so I take it from the other comments here that Kent isn’t entirely correct? Is there a summary of the back and forth so far?

    • unknownuserunknownlocation@kbin.earth
      link
      fedilink
      arrow-up
      10
      ·
      12 hours ago

      I might not have everything, but here’s the best summary I can put together:

      This back and forth has been going on for a while now. The main complaints more recently have been the timing of his pull requests, and just generally his attitude and cooperation with others. The most recent spat was because he submitted a feature in the rc3 merge window, whereas you’re only supposed to submit Bugfixes in that time frame. The feature in question was a journal rewind function, which would essentially move the filesystem back in time, which could fix an issue that did crop up in the testing phase. As such, he saw it as a workaround to fix an issue that had arisen, and so despite it technically being a feature, he saw it in the category of Bugfixes. The caused major disagreements as well as the way he talked with others. And now his pull request for rc1 has been simply ignored by Linus.

      The point where Kent is coming from is that he wants a rock solid file system, and he’s following a bit of a take no prisoners approach to reach that goal. He seems to get most of his income from his following on Patreon, and so his focus is squarely on the users. With that focus, he seems to lose sight of other things, especially the cooperation with others in the kernel team. In fact, a number of people he has sparred with have shown decent respect for his code recently, saying the problem is really the cooperative aspect. One of the main reasons for bcachefs is also the lack of a proper CoW-filesystem in the Linux kernel that doesn’t have the kinds of problems that btrfs has. And the fact that he states this and also talks about the lessons he’s learnt from btrfs’s shortcomings rubs a number of people the wrong way.

      Now here’s some stuff I read into this personally: I have the impression that Kent looks up to Linus in a way. And they’re actually both kind of similar: they both are extremely talented engineers, they both saw something missing in the software landscape and said “fuck it, I’ll make it myself”, and they both can be pretty serious dicks. I mean Linus managed to get suspended from his own damn project for being a dick - now that’s an achievement. He’s older now and somewhat calmer, but even recently he had quite the outburst on the mailing list. And I get the impression I get is that Kent (probably subconsciously?) has an attitude of “if he can do it, so can I”. Which would be fair (even though it is poisonous), the only problem being Linus having the longer lever.

      Then there’s the aspect of his mental health. He has said multiple times that his mental health has been suffering, which honestly doesn’t surprise me. And if you look at his responses in different places, there seems to be quite an up and down. In some cases, he’s very respectful to Linus, and in some cases he’s pretty nasty (yes, Linus level nasty, but still). As far as I can tell, he needs a break and therapy. The only problem being, bcachefs has quite the momentum currently, and it wouldn’t exactly be great for the project to lose that momentum, either. (Mind you, probably still better than being kicked from the kernel)