Package: release.debian.org Severity: wishlist Dear Release Managers:
Release Policy here https://release.debian.org/stretch/rc_policy.txt says: "Packages must autobuild without failure on all architectures on which they are supported". Please clarify how should this policy be applied when the package FTBFS randomly. My interpretation of Release Policy is that random FTFTS bugs are also a kind of failure and the bug is therefore severity:serious, like packages which FTBFS all the time. However, if this is a wrong interpretation, then please write something about it in Release Policy regarding FTBFS bugs which happen randomly, as we are still discussing about these kind of things over and over again and not everybody seems to interpret Release Policy in the same way. In particular, I would like you to review the following bugs and determine if it's justified enough to downgrade them and what would be the right severity for them: -------------------------------------------------------- https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=837067 Rationale for downgrading: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=837067;msg=14 None provided. -------------------------------------------------------- https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=841098 Rationale for downgrading: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=841098#78 "The maintainer does not intend to work on the issue". -------------------------------------------------------- https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=843038 Rationale for downgrading: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=843038#44 "Not enough CPU power to pass the tests". -------------------------------------------------------- While we are at it, there is also some controversy about the meaning of "autobuild". Some people sometimes downgrade FTBFS bugs with the only rationale that they "do not happen in buildd.debian.org". My interpretation of "autobuild" here is that it must build ok not only in buildd.debian.org but also in any other similarly-configured autobuilder as far as it follows the common standards to build packages. This is, I believe, consistent with the fact that we consider a missing build-depends as RC even if the missing package is still installed by default in the official autobuilders. OTOH, if my interpretation is not correct, and the "must" (i.e. the RC-ness of the bug) is only for packages building in buildd.debian.org, please explain how should we apply "must autobuild" when package is Arch:all and the maintainer uploads all the .debs. Thanks.
