Bug#679746: [Adduser-devel] Bug#679746: Bug#679746: default --system home directory leads to piuparts RC bugs
On Thu, Dec 13, 2012 at 07:17:47AM +, Stephen Gran wrote: This one time, at band camp, Andreas Barth said: I suggest to tolerate the current behaviour for the upcoming release. Afterwards, I suggest to either require either --no-home or --home for system users. Or switch defaults to another location, e.g. /var/lib/syshome (or whatever else). Ew. That seems to me to be inventing something ad-hoc and Debian-specific, which feels wrong. adduser is already Debian specific. It is a wrapper around useradd which does things right from a Debian policy point of view, making things easier for a package maintainer by making it harder to make mistakes in postinst. Currently, packages become RC-buggy for just adding system users without --no-home and no --home (even if not relying on the directory). I think that - if we read policy as that - then it's better to fail the postinst then to have hidden RC bugs. Explicit RC bugs are always better then well hidden ones. (Of course, all of that for after this cycle.) So we're not going to do an MBF, but we're going to make code changes so that packages blow up at install time and users do the MBF for us, one by one? How bizarre. That can't be what you're actually saying. piuparts is already doing those MBFs. Maybe we don't make packages blow up at install times first but we can print a warning. Greetings Marc -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#679746: [Adduser-devel] Bug#679746: Bug#679746: default --system home directory leads to piuparts RC bugs
On Thu, Dec 13, 2012 at 07:17:47AM +, Stephen Gran wrote: This one time, at band camp, Andreas Barth said: Currently, packages become RC-buggy for just adding system users without --no-home and no --home (even if not relying on the directory). I think that - if we read policy as that - then it's better to fail the postinst then to have hidden RC bugs. Explicit RC bugs are always better then well hidden ones. (Of course, all of that for after this cycle.) So we're not going to do an MBF, but we're going to make code changes so that packages blow up at install time and users do the MBF for us, one by one? How bizarre. That can't be what you're actually saying. A first step could be to have adduser --system spew a warning in this case. Maintainers should notice that when testing packages before upload. Additionally, there could be a lintian check for that. This course of action has numerous precedences. Greetings Marc -- - Marc Haber | I don't trust Computers. They | Mailadresse im Header Mannheim, Germany | lose things.Winona Ryder | Fon: *49 621 31958061 Nordisch by Nature | How to make an American Quilt | Fax: *49 621 31958062 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#679746: [Adduser-devel] Bug#679746: Bug#679746: default --system home directory leads to piuparts RC bugs
This one time, at band camp, Andreas Barth said: * Stephen Gran (sg...@debian.org) [121212 08:19]: This one time, at band camp, Andreas Barth said: Now, I don't agree that this is a bug in adduser - I think this is a bug in packages invoking adduser. I think the right way to do this is a mass bug filing on those packages. I think I disagree here. If we have a switch on adduser --system saying that the user to be created is a system-user, I tend to think that adduser should then select an appropriate place to create $HOME. (I wouldn't mind if we say that for the upcoming release, this still is within /home, and we fix that for the next stable release.) Where would you suggest? I suggest to tolerate the current behaviour for the upcoming release. Afterwards, I suggest to either require either --no-home or --home for system users. Or switch defaults to another location, e.g. /var/lib/syshome (or whatever else). Ew. That seems to me to be inventing something ad-hoc and Debian-specific, which feels wrong. I can't think of a better place, in the absence of direction, than /home. If packages rely on the presence of a home directory, it's up to them to ask for a sensible one. Currently, packages become RC-buggy for just adding system users without --no-home and no --home (even if not relying on the directory). I think that - if we read policy as that - then it's better to fail the postinst then to have hidden RC bugs. Explicit RC bugs are always better then well hidden ones. (Of course, all of that for after this cycle.) So we're not going to do an MBF, but we're going to make code changes so that packages blow up at install time and users do the MBF for us, one by one? How bizarre. That can't be what you're actually saying. Cheers, -- - | ,''`.Stephen Gran | | : :' :sg...@debian.org | | `. `'Debian user, admin, and developer | |`- http://www.debian.org | - signature.asc Description: Digital signature