Re: primary.xml.gz 'file' questions

> Is there a reason why certain files and directories appear in both
> xml.gz's and some just appear in filelists.xml.gz?

Metadata size. Yum downloads filelists on demand, and we try to avoid it.
IMO "Requires: /filelists-only-file" should be considered a packaging 
error and filelists should be dropped completely.

> Is there a special set of RPM attributes (from
> http://rpm.org/api/4.11.1/rpmfi_8h.html possibly?) that I could
> examine to determine whether a file/directory should appear in
> primary.xml.gz ?

It's based on the filename.  Everything that starts with /etc/
or contains bin/ is included in primary.

Look at re_primary_filename() in 
/usr/lib/python2.7/site-packages/yum/misc.py