IRC channel logs
2026-05-27.log
back to list of logs
<sam_>rsmarples: interesting :) <sam_>it's still very much in its infancy so i'm glad it has proven useful already <solid_black>okay, maybe this works? polari is badly broken, and kiwiirc fails to connect <jab>solid_black: welcome back! <jab>solid_black: the coolest hurd news is that we have some unmerged "ext3/4fs" patches. I'm using them on real hardware. <jab>It's just pending review. <solid_black>I've been low-key following the mainling list, but not really looking into details <solid_black>is it real ext3/ext4, or I think there were some patches to add a custom journal to ext2? <sam_>bc8ac23d670ad3aeaddf693d30cdc2abbf9952af landed earlier :) <sam_>solid_black: I think the question is somewhat academic <sam_>if it's not literally the fs/ext4 implementation from linux.. <sam_>it's been great fun and will continue to be I'm sure <solid_black>I mean, my understanding was that the journal format was custom, and not in any way ext[34]-compatible <sam_>i think they said they even replayed it on linux <solid_black>I was, at the time, thinking we should have an all-new ext4fs based on the ext userspace libraries from linux <sam_>I posted a few weeks ago a kind of funny thing wrt plans for ext4 to be runnable via fuse <sam_>I don't but I'm new to following Hurd, I did see some bits about it when researching though <sam_>I was curious about it but didn't ask more yet, because we have something that relies on fuse that would be nice to have <solid_black>a real libfuse for the Hurd that had an API compatible to that of Linux's libfuse, but internally structured like the other Hurd translator frameworks (libdiskfs, libnetfs) <jab>I'm running and64 on a T420. No issues. <jab>with I think 12 or 16 GB of ram. <solid_black>I saw there was some talk about pageout and deadlocks recently, I think? <solid_black>would be great to page (ha!) that back in, and help out <bjc>am i reading this right? did the ext2fs journaling stuff get committed? <jab>bjc: it is just waiting review. <jab>I am running it on real hardware. <sam_>[15:16] <sam_> bc8ac23d670ad3aeaddf693d30cdc2abbf9952af landed earlier :) <jab>I can send you the email link on how you can run it yourself. <jab>if you want to be a tester of it. <bjc>jab: see the commit sam_ posted <jab>I'll take a look after lunch. <jab>off topic, but if ya'll want some fun videos to watch on your lunch break today...I've really been getting into the fat electrician. He does american military history videos that are hilarious! <solid_black>I found my old work on libfuse, and also on reworked bootstrap <solid_black>hopefully someone can make sense of it, perhaps even myself <jab>thanks for committing it somewhere. :) <jab>might not be a bad idea to send an email to bug-hurd about it too. <sam_>samuel already replied to the thread saying it <solid_black>I remember that I quite disliked something about how the existing libfuse (which was a result of a gsoc project I believe?) was implemented <solid_black>it could be that it was hooked up to netfs, or something like that? I don't remember the specifics <solid_black>I strongly felt that it should be done differently, and started my own version <azert>how does your libfuse differs deom <jab>soo cool to see "ext3/4fs" patches merged. Maybe we should start calling it ext3fs... <azert>solid_black: oh I see.. seems like a great idea! Basically a native libfuse for the Hurd! <azert>I always wondered how filesystems that are not diskfs based handle mmap and paging