On 6 Apr 2016 22:12, "Tom Horsley" <horsley1953@gmail.com> wrote:
>
> On Wed, 06 Apr 2016 12:55:04 -0700
> Doug H. wrote:
>
> > So my guess is that there is a digest file that is being checked by
> > default in fedora but not checked on centos.
>
> But I'd really like to know why it says it has a problem.
> We use rpmbuild on a fairly old RHEL build system to create
> the packages (so the programs will be backward compatible
> with older releases) and we've never had a problem before.
>
> And why is it only complaining about one file? And one
> package? (All the rpms I installed were built on the
> same build system, and just this one rpm has an install
> problem).
>
> Is the python digest code incompatible with the C code
> in rpmbuild in one out of 10,000,000 cases or something?

Incidentally this is not the best way round of doing this. Bear in mind the builders run on Fedora.current in koji and yet we build epel5 there fine.

Using mock is always the sensible thing to do rather that rpmbuild directly and anything built using that should work on the target system - if not it's a bug and it'd break koji builds too.