Hi,
On Wed, Jul 15, 2026 at 08:31:54PM +0200, John Paul Adrian Glaubitz wrote:
On Wed, 2026-07-15 at 18:23 +0200, Cyril Brulebois wrote:
Starting from a clean Git checkout (meaning reallyclean is a no-op),
the build_netboot target gives me a build/dest/netboot/mini.iso that
lets me start and finish partman successfully.
All I can say is that I ran into this circular dependency issue as well
when building d-i locally with the udebs partman-hfs and hfsprogs-udeb
injected manually and it only went away after I dropped partman-base from
Depends of partman-hfs.
That's the current state of partman-hfs's master branch. I have rebuilt
partman-hfs as well (since the version in the archive still depends on
hfs-modules, hfsprogs-udeb, partman-base, partman-basicfilesystems,
cdebconf-udeb), my locally built partman-hfs just has
partman-basicfilesystems, cdebconf-udeb.
My circular dependency abort remains.
I was building inside a freshly generated schroot, so the build environment
was clean in my case.
So you built a clean unstable schroot, and did the entire mr checkout in
the chroot? That would mean that some package that might be installed on
my daily driver breaks the build?
Since I did not change anything but updating the d-i
repository with "git pull", I assume this to be a regression in d-i itself
or at least a requirement to implement some changes to udebs if dropped to
the "localudebs" directory.
Maybe it's just partman as being the most complex part of the installer?
Greetings
Marc
--
-----------------------------------------------------------------------------
Marc Haber | "I don't trust Computers. They | Mailadresse im Header
Leimen, Germany | lose things." Winona Ryder | Fon: *49 6224 1600402
Nordisch by Nature | How to make an American Quilt | Fax: *49 6224 1600421