Reject or strip Unicode control characters?

[dralley]

Unicode control characters in metadata cause downstream issues, particularly with regards to the XML metadata that is eventually generated when making a repository out of these RPMs.

rpm-software-management/rpm#2742
rpm-software-management/createrepo_c#327

We should consider sanitizing inputs or rejecting invalid ones that we know have the potential to cause problems. Of course, with the focus of this package being a library rather than a build tool, I can also see an argument that this logic belongs elsewhere (e.g. a consumer of the library). I'm of mixed opinion at the moment but it's worth a discussion.

[dralley]

Example: rpm names have a restricted character set.