IRC channel logs

<AidenIsik>Hi, I've run into an issue trying to boot (Debian) Hurd on bare metal. It gets up to disk initialisation, then I get `piixide1:0:0: lost interrupt` every ten seconds.

<AidenIsik>From what I understand this is rumpdisk, as searching this error brought up results from NetBSD

<AidenIsik>Is there a way I can disable rumpdisk for now and fall back to the in-kernel IDE driver?

<AidenIsik>(Running on a Corebooted system btw)

<damo22>sneek: later tell AidenIsik if possible, you need to change your nvram settings to sata = AHCI mode then it should just work

<sneek>Will do.

<AidenIsik>damo22: sneek: Thanks for the info

<sneek>AidenIsik, you have 1 message!

<sneek>AidenIsik, damo22 says: if possible, you need to change your nvram settings to sata = AHCI mode then it should just work

<AidenIsik>I was reading the IRC logs

<AidenIsik>I will try that

<hwpplayer1>Anyone developing Hurd on Mac Silicon ? Which virtualization software ?

<hwpplayer1>I will try remote development thanks

<hwpplayer1>in a datacenter

<solid_black>hello :)

<hwpplayer1>hello solid_black

<solid_black>hey hwpplayer1

<solid_black>azert: Mach IPC already has an epoll-like feature, namely port sets

<hwpplayer1>what's going on there solid_black *

<hwpplayer1>?

<solid_black>that is enough to implement various multiplexing when using Mach IPC directly

<solid_black>but not to implement specific Unix-level epoll semantics

<solid_black>in particular it has to be a file descriptor, so there should be a server implementing it

<solid_black>hwpplayer1: I'm at a BoF (chat) for local-first technologies in GNOME (mostly based on p2panda) at GUADEC 2026 :)

<hwpplayer1>where ?

<solid_black>in A Coruna, Spain

<hwpplayer1>spain

<hwpplayer1>yes

<hwpplayer1>I like Spain :)

<hwpplayer1> https://events.gnome.org/event/306/

<solid_black>that's the one, yes

<solid_black>I gave a talk on the first day, about my work at $dayjob over the few previous years

<solid_black>and led a workshop today just recently

<solid_black>where we built a simple app for drawing (or taking notes) with a stylus wih GTK, C++, and peel

<hwpplayer1>yes I like C++

<hwpplayer1>which standard ?

<solid_black>peel supports anything upwards of C++11

<hwpplayer1>I see I learn beginning from 17

<hwpplayer1>but I will read older standards

<hwpplayer1>because I have to

<hwpplayer1>:)

<solid_black>it is largely the same language, just with less or more obscure features

<hwpplayer1>I can imagine

<hwpplayer1>Where are you from ? nationality ?

<hwpplayer1>Turkish/Albaian/Ex-Yugoslavian

<hwpplayer1>Albanian

<solid_black>heh, guess I don't have a real name set here?

<hwpplayer1>Sergey Bugaev ?

<solid_black>yes, so I do have it set then

<hwpplayer1>I should have been know you I might have been heard

<hwpplayer1>Always when I ask someone's name on IRC I feel the Matrix scene meeting neo and trinity :D

<hwpplayer1>My name is Mert Gör

<hwpplayer1>Living in İstanbul

<hwpplayer1>I try to be a systems programmer

<solid_black>oh cool, I'd like to properly visit Turkey some time

<hwpplayer1>where in Russia ?

<hwpplayer1>I heard Moscow State University

<hwpplayer1>searched for it

<solid_black>MSU is where I studied

<hwpplayer1>I see

<hwpplayer1>did you studied a master degree or more ?

<hwpplayer1>I graduated economy science but I want CS

<solid_black>I quit after master's, since I was quite disappointed with the higher education system (at least at the time, in Russia)

<hwpplayer1>why ?

<hwpplayer1>why disappointed I mean

<solid_black>it was... pointless, not useful, just a lot of... verbiage

<solid_black>none of it was really interesting

<solid_black>whereas at a company, or in a free project, I could do some actually interesting cool things

<solid_black>and have long been doing, by that point

<solid_black>hacking on the Hurd included

<hwpplayer1>I see

<hwpplayer1>"hacking" is more important yes

<hwpplayer1>to practice

<hwpplayer1>not only theories

<solid_black>but, since last year, I have myself been teaching seminars on like C and AArch64 and Unix at the HSE / CU

<hwpplayer1>Do you have a git account ?

<hwpplayer1>which I can follow

<solid_black>git doesn't have accounts

<hwpplayer1>like GitHub I mean

<hwpplayer1>Codeberg, Sourcehut etc

<solid_black>on GitHub, https://github.com/bugaevc, but there's not much stuff happenning there these days

<hwpplayer1>I see

<solid_black>peel is on GNOME GitLab, but also there's a mirror on GitHub

<hwpplayer1> https://github.com/hwpplayer1

<hwpplayer1>What is your suggestion for Microkernel architecture while learning ?

<hwpplayer1>solid_black: ^

<hwpplayer1>do you know GNU/Hurd licensing ?

<hwpplayer1>I contribute to Free Software Directory every Friday

<hwpplayer1>maybe we can work on that together

<azert>solid_black: if it is just to implement a file descriptor out of a port set, I think glibc can do that.

<azert>I really suspect that the reason you want a server is such that the file descriptor outlive the process that creates it, but is this really required?

<azert>what I mean is that glibc could create a thread and serve the file descriptor for other processes. This is even cleaner from a resource management perspective I think

<solid_black>I don't particularly care what happens with the epoll after the process dies, though technically we should support that

<azert>exactly

<solid_black>but all real uses of epoll have you associate either the fd number (which is only meaningful inside the process) or a pointer (ditto) with an interest

<solid_black>you could try to move the epoll server to live in-process on a different thread, but I don't see much of an advantage

<solid_black>an example thing that is used by real life programs and must work: adding a epoll fd to another epoll (or poll/select)

<azert>could this be implemented with port sets?

<azert>the reason for the epoll server living in a thread in the creating process is just because it’s more Hurd-like

<azert>the reason to move such a function in gnumach is just because it is probably something useful

<hwpplayer1>ping

IRC channel logs

2026-07-19.log