On Thu, Aug 08, 2019 at 09:43:53AM +0200, Emmanuel Bourg wrote: > If a build issue appeared *after* the release, due to a regression > caused by a point release, then yes I think the issue is worth reporting > (but I wouldn't bother reporting minor test failures that can be easily > skipped to rebuild from sources).
You keep saying "minor" but it actually makes the build to fail! This is certainly not what we expect to happen in a supported release and it's already written in Release Policy: Packages must build without failure, it refers to dpkg-buildpackage exit status, which in this case indicates failure, it does not refer to the build *without* the tests. And no, it's not easy to skip, by default DEB_BUILD_OPTIONS does not contain nocheck, so by the time the end user realizes that the package fails to build, it's too late and we have already make them to lose their time, because the process of building the package stopped and the binary packages are not generated. Point releases exist to fix bugs like this one, so I don't see why I would have to stop reporting them. In either case, I'm not sure where is the problem for you. I'm not specifically asking you to do anything (I could not, we are all volunteers), and there is no autoremoving of packages in stable or oldstable. So you could ignore the bug if you dislike it so much. I only ask that you respect that this is a serious bug in stretch, which means keeping the door open for a fix, i.e. allowing somebody else to fix if you are unwilling. That's all. Thanks.