On Wed, Mar 09, 2005 at 01:23:29PM -0500, Jeff Spaleta wrote:
While I don't think i would actively encourage people to work on
this.. from a manhour priority perspective... if interested parties
are intent on spending their personal time on this.. I would say there
is only one way forward..and that is incrementally. Instead of
fighting to convince developers for existing projects to include new
functionality they have reserverations about... build a stand-alone
tool set that can be used to generate the deltas and then re-generate
the rpms on the clientside for the distro package management tools to
use. Find a subset of mirrors to offer that service and get people to
test it. First get a stand-alone implementation tha can be used to
layer the experimental functionality over existing tools... before
worrying about integrating that functionality into existing tools.
That's a good idea. I'd start here:
http://zsync.moria.org.uk/
Yep, client-side rsync. I don't know the patent status in the US
(Google for earlier discussions), and don't want to know. Get it
before the EU becomes a police state too. :-(
It should be an afternoon's work to combine the above with:
http://rpmrebuild.sourceforge.net/
and have a tool that will mirror update directories that contains
.zsync files. Run some tests locally, then convince a repository
in the free world to give it whirl. If it works out, perhaps the
signatures could go in a repository metadata file.
Regards,
Bill Rugolsky