Hello release team, Christian Kastner discovered that when using vmdebootstrap with autopkgtest's "VM setup" script it erroneously enables dpkg's "force-unsafe-io" option on the host instead of the guest. This can lead to potential system damage and thus should qualify as RC:
https://bugs.debian.org/775076 This is fixed in version 3.9.3 in unstable now, including a postinst cleanup for cleaning up an erroneously created /etc/dpkg/dpkg.cfg.d/autopkgtest on the host. At the time when testing froze, testing had 3.6, and unstable's 3.7 (with a lot of changes) was one day short of propagating to testing; so as these were already out of sync, I just kept on uploading to unstable as that's what most people use for development. So I prepared a debdiff for testing-proposed-updates at https://bugs.debian.org/cgi-bin/bugreport.cgi?msg=34;filename=autopkgtest.775076.jessie.debdiff;att=1;bug=775076 autopkgtest is mostly a Debian/Ubuntu developer, QA, and continuous integration tool, so honestly the testing version isn't that interesting any more. So if you prefer, I'd be perfectly okay (and in fact happier) with getting 3.9.3 into testing instead; it has a lot of new features, but also a lot of bug fixes: https://people.debian.org/~mpitt/tmp/autopkgtest_3.6_3.9.3.debdiff It's clearly waaaay outside the current freeze rules, but as it's not an end user tool but rather an internal QA one, I want to at least consider it. If we want to go that route, I'm happy to provide more details. Thank you and have a nice Sunday! Martin -- Martin Pitt | http://www.piware.de Ubuntu Developer (www.ubuntu.com) | Debian Developer (www.debian.org)
signature.asc
Description: Digital signature
