João, le dim. 06 avril 2025 20:27:44 +0200, a ecrit: > On Sun, Apr 06, 2025 at 07:33:59PM +0200, Samuel Thibault wrote: > > João, le ven. 28 mars 2025 17:39:17 +0100, a ecrit: > > > I have not yet tried to get the hurd-console working, but this will > > > probably > > > involve starting the console server (under runit supervision?), setting > > > the > > > translators for the different tty and running getty on them? > > > > That is essentially it, yes. The console server would not be under > > runit supervision: it's just sitting on /dev/console. It's the client > > that would be run by runit.
Sorry, I confused things. /dev/console is the mach console indeed. The latter is where the console client exposes some information. It's /dev/vcs on which the console server sits. > You mention /dev/console, is this where the getty should > attach instead of /dev/tty[n] when using hurd-console? No, it's indeed /dev/tty[n] on which the console server throws the content. > I thought the problem could be that the console translator is not setup > correctly. Is this done at some stage during the boot process? How? Or is it a > passive translator set at install time? See console/README.UTF8, /dev/vcs is a passive translator, see MAKEDEV's rule to set it. > > > On the topic of init systems, many will know that GNU Shepher 1.0 has been > > > released > > > (https://www.gnu.org/software/shepherd/news/2024/12/the-shepherd-1.0.0-released/), > > > and it is in Debian but not yet building on the hurd > > > (https://buildd.debian.org/status/package.php?p=shepherd). > > About Shepherd, guille-3.0 is a dependency, and that failed to build with an > odd log: > > df -h # See what we have (relevant to tests like ports.test) > df: cannot read table of mounted file systems: No such file or directory debootstrap is apparently dropping /etc/mtab, I have put it back, it's building. Samuel
