IRC channel logs


back to list of logs

<kitty1>yo, anyone know if anything of note has happened since the last time guix had a hurd-themed blogpost?
<z3ntu[m]>Are chroots supported on Hurd? I know they exist and generally do work but I often hit super weird problems with them
<youpi>they do, all packages are built inside them
<youpi>note however that you need to bind-mount /dev and /servers to avoid triggering duplicate translators
<z3ntu[m]>e.g. `make: /tools/bin/mkdir: Computer bought the farm` when trying to build diffutils
<z3ntu[m]>I do have these bind commands, are they ok? `settrans -kp ~/myroot/servers /hurd/firmlink /servers` and `settrans -kp ~/myroot/dev /hurd/firmlink /dev`
<youpi>ah, that reminds me a bug in libc that was fixed in spawn
<youpi>which box are you using?
<z3ntu[m]>host is latest debian hurd 11 image, I think 202108something
<z3ntu[m]>"guest" (tools in chroot) is a concotion from July last year
<youpi>which version of make and libc0.3 does it have?
<youpi>(in the chroot)
<z3ntu[m]>GNU Make 4.3 and GNU C Library (GNU libc) development release version 2.33.9000.
<youpi>please show the debian package versions
<z3ntu[m]>chroot is not debian
<youpi>they have a lot of importance in terms of binary rebuilds and patches
<youpi>so that must be the same bug we have seen in debian with make rebuilt with a recent glibc
<z3ntu[m]>it's definitely built with that version of glibc, there was no rebuilding happening with this rootfs
<youpi>you want this upstream glib cpatch
<youpi>hurd spawni: Fix reauthenticating closed fds
<z3ntu[m]>ok will try to build a new rootfs with that
<z3ntu[m]>random question, any clue what this build error is about? gcc-11.2.0.obj/i686-pc-gnu/libstdc++-v3/include/fenv.h:58:11: error: 'fenv_t' has not been declared in '::'
<youpi>I'd need much more context
<youpi>with just that, I'd think that there's a C / C++ mismatch
<z3ntu[m]>I guess either caused by new hurd/mig/gnumach or new glibc, because the rest of the versions here were built before