On Wed, Feb 9, 2011 at 1:45 AM, Toshio Kuratomi <a.badger(a)gmail.com> wrote:
> > Yeah, I wish I could find a good cite...
> > Perhaps the deltarpm and/or xz maintainers could chime in with the
> > exact info.
>
> This is the first hit for "xz compression deltarpm problem" and
> describes the problem perfectly ... although it is from 2009 ;)
>
>
http://cedarandthistle.wordpress.com/2009/12/29/deltarpm-problems-part-ii/
>
<nod> That does a great job of explaining what we had here. The difference
this time is that instead of architecture being the key, version of xz was
the key. The new and old versions of xz create compatible archives but they
aren't byte-for-byte identical. That could lead to certain circumstances
where the rpm signature was invalid for an rpm reconstructed from a deltarpm
even though the uncompressed data was the same.
I updated the wiki page a bit.
josh