[Piuparts-devel] Bug#766921: piuparts fails with "ValueError: too many values to unpack" when dpkg prints a warning

[Andreas Beckmann]

  I: Found additional required dependencies: acl adduser dmsetup
init-system-helpers insserv libaudit-common libaudit1 libbz2-1.0 libcap2
libcap2-bin libcryptsetup4 libdb5.3 libdebconfclient0 libdevmapper1.02.1
libgcrypt20 libgdbm3 libgpg-error0 libkmod2 libncursesw5 libprocps3
libsemanage-common libsemanage1 libslang2 libsystemd0 libudev1
libustr-1.0-1 perl perl-modules procps systemd systemd-sysv udev

Why does a minimal chroot contain systemd?
A systemd environment may get funny for --allow-database :-)

  I: Found additional base dependencies: binutils bzip2
debian-archive-keyring dpkg-dev gnupg gpgv libapt-pkg4.12 libdpkg-perl
libeatmydata1 libreadline6 libstdc++6 libtimedate-perl libusb-0.1-4 make
patch readline-common xz-utils

So something pulls in dpkg-dev, but not gcc. That sounds wrong. Both.
dpkg-dev shouldn't be needed in a minimal chroot (w/o build-essential).

2m30.4s DEBUG: Starting command: ['chroot', '/tmp/tmpMtfEHx',
'eatmydata', 'mount', '-t', 'proc', 'proc', '/proc']
2m30.4s DUMP:
  sh: 1: gcc: not found
  dpkg-architecture: warning: couldn't determine gcc system type,
falling back to default (native compilation)
2m30.4s DEBUG: Command ok: ['chroot', '/tmp/tmpMtfEHx', 'eatmydata',
'mount', '-t', 'proc', 'proc', '/proc']

Whatever calls dpkg-architecture for running mount ...

Does the same problem happen if eatmydata is *not* used? (That's the
only factor piuparts may contribute to the issue.)
Try --no-eatmydata

Andreas