Well, that's fuckin' awesome.  Someone could tag Openoffice to use the yum plugin if they wanted it to be updated using the deltas approach and save 100 MB just like that.

Implementing the delta updates with the yum plugin is perfect.  I'm going to follow this up and see what I can learn.  I hope there are others out there as well that are interested in working on this.  It's not my idea, but definitely needed your comments below to move this idea forward.

If a few different methods were available for updating rpms then there could be another level that automates the decision making to look for the most feasible method to use.

Benjy,
AWWTF



On 11/29/05, Jeff Spaleta <jspaleta@gmail.com> wrote:
On 11/28/05, Benjy Grogan <benjy.grogan@gmail.com> wrote:
> This must be done.  Who wants to start?  Who could start?
Must? You're mastery of hyperbole takes my breath away.

It's already been suggested to you that you look back into the
archives the the this list and find previous discussions. You seem to
be suffering from the delusion that this is a new idea. This comes up
every release.. and yet it doesn't get much further than a proof of
concept code release that gains absolutely no traction.

Since you seem incapable of searching the archives on your own let me
point out a previous discussion:
https://www.redhat.com/archives/fedora-devel-list/2005-March/msg00344.html

Please, read the whole thread and take the time to figure out the
people in the thread who agree with you as to what is needed, contact
those people and see how far they have gotten with a usable
implementation.

Hell, now that yum's plugin support is exposed there is absolutely no
reason stopping people who are interested in showing this is
worthwhile to support officially from cobbling together a yum plugin
to put into extras and persuading a set of mirrors to keep the daltas.
With plugin support in yum, the complexity of the proxy server in the
thread i referenced above is problably no longer needed.

Instead of more exaggerated rhetoric, what's need to move this concept
forward is a usable implementation that can go into Extras and a set
mirrors willing to support your delta approach.  And then find a way
to keep metrics as to how popular the delta process is and whether its
worth the upkeep to maintain.

-jef"mother-smurfer"spaleta

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list