Ownership of /usr/lib/rpm/fileattrs
ppisar at redhat.com
Mon Oct 19 12:09:57 UTC 2015
On 2015-10-16, Orion Poplawski <orion at cora.nwra.com> wrote:
> only rpm-mpi-hooks requires rpm-build for directory ownership, while
> javapackages-tools takes the route of owning the directory. However, I'd
> rather rpm-mpi-hooks not require rpm-build as it's not really necessary other
> than for this directory. The simple thing I think would be for rpm to own the
> directory. Does that seem reasonable?
If you move the directory from rpm-build to rpm, then all the packages
should require rpm instead of rpm-build.
But that contradicts your wish "I'd rather rpm-mpi-hooks not require
rpm-build as it's not really necessary other than for this directory".
rpm-mpi-hooks does not need rpm more or less than it needs rpm-build,
If the only reason for the dependency is the ownership, then co-owning
the directory as javapackages-tools does is the right way.
But the directory hosts dependency generators executed by rpmbuild. Not
by rpm. Moving the ownership from rpm-build to rpm sounds semantically
One can perceive the package packages as rpm-build plugins. Having the
dependency on rpm-build does not look wrong in the end.
More information about the devel