Request for Comments: updating RPMs using binary deltas

[Lamar Owen]

On Friday 09 January 2004 12:02 am, seth vidal wrote:
> > But I ask again: if SuSE is doing this, is it working for them and their
> > users RIGHT NOW?  I don't have the answer; I just threw out the idea. 
> > The few users that have posted in fedora-list have seemed enthusiastic.

> no way, enthusiatic users on fedora-list who won't have to deal with the
> consequences if such a thing were done? I'm shocked to hear that.

IMO the consequences are minimal for mirror operators.  Developers of update 
tools (like yourself) would have more of a job; and that is probably why you 
are in principle opposed since it would create more work for you as a 
developer.  But not as a mirror op.  It would be a one-time pain to develop 
the algorithm.  The build hosts of course have to generate the deltas.  This 
may or may not be difficult.  I see a single program doing the job that gets 
run post build on the two trees (much like a diff -r) and generates the tree 
of deltas.  This same tool could even be used by mirrors to pull only the 
deltas and locally regenerate the larger full updates.  There is, however, 
the signature issue there, so that might not be easily workable.

> And the response to the Suse^WNovell patching mechanism by a variety of
> 'authorities' on rpm has been "yeeeeeeeeech"

Who?  When? What forum?  I'm not thrilled by some of the details, but the 
concept is a sound one.  However, I would like to read that stuff, and I 
don't recall seeing it come down rpm-list.  And I don't do IRC.

> so if you feel like you're getting pushback it's b/c there is a lot of
> reaction to things that will create an enormous amount of complexity,
> cost a lot of development time and bear out questionable fruit.

You have yet to prove that it would create an _enormous_ amount of complexity 
or cost _a_lot_ of development time.  But the bandwidth fruit for the end 
user is anything but questionable, if we can get better patch to noise ratio 
than SuSE is getting, which is already pretty good.

So your view is that your time is more valuable than the end user's time and 
frustration, particularly if they have poor ISPs.  That is how it sounds, 
whether that is your intention or not.  I can see users that will use SuSE 
simply because they do this.
-- 
Lamar Owen
Director of Information Technology
Pisgah Astronomical Research Institute
1 PARI Drive
Rosman, NC  28772
(828)862-5554
www.pari.edu