Hi, I'm just looking at the rpmlint errors for the xpra package, and one I am stuck on is this: xpra.x86_64: E: dir-or-file-in-var-run /var/run/xpra The package drops a file in /usr/lib/tmpfiles.d which causes the creation of /var/run/xpra when needed. The package has this in %{files}: %ghost %dir %{_localstatedir}/run/xpra which seems reasonable to me. So I don't really understand the rpmlint error. I picked two packages at random, fail2ban and systemd, which also install directories under /var/run and have those directories listed as %ghost. So, is the rpmlint error incorrect? Or, should these guidelines: http://fedoraproject.org/wiki/Packaging:Tmpfiles.d also apply to /var/run? In which case, many packages need fixing to use %verify, I think.

On 2 October 2016 at 04:46, Orion Poplawski <orion(a)cora.nwra.com&gt; wrote: > On 10/01/2016 02:00 PM, Jonathan Underwood wrote: >> >> Hi, >> >> I'm just looking at the rpmlint errors for the xpra package, and one I >> am stuck on is this: >> >> xpra.x86_64: E: dir-or-file-in-var-run /var/run/xpra >> >> The package drops a file in /usr/lib/tmpfiles.d which causes the >> creation of /var/run/xpra when needed. >> >> The package has this in %{files}: >> >> %ghost %dir %{_localstatedir}/run/xpra >> >> which seems reasonable to me. So I don't really understand the rpmlint >> error. I picked two packages at random, fail2ban and systemd, which >> also install directories under /var/run and have those directories >> listed as %ghost. >> >> So, is the rpmlint error incorrect? Or, should these guidelines: >> >> http://fedoraproject.org/wiki/Packaging:Tmpfiles.d >> >> also apply to /var/run? In which case, many packages need fixing to >> use %verify, I think. > > > # ls -l /var/run > lrwxrwxrwx. 1 root root 6 Feb 16 2016 /var/run -> ../run > > /var/run is the same thing as /run, so yes they apply. Not sure if we > really need *every* file/dir owned by a package - especially ones in tmpfs > filesystems. Right. So, the rpmlint error directly contradicts the tmpfiles.d packaging guidelines then. I'll file a bug against rpmlint.