[fab] Re: Firefox splash page tracker

seth vidal skvidal at linux.duke.edu
Fri Oct 6 13:42:02 UTC 2006


On Fri, 2006-10-06 at 09:35 -0400, Christopher Blizzard wrote:
> Mike McGrath wrote:
> > On 10/4/06, Greg DeKoenigsberg <gdk at redhat.com> wrote:
> >>
> >> On Wed, 4 Oct 2006, Robert 'Bob' Jensen wrote:
> >>
> >> > I hope you are understanding it correctly because that is how I
> >> > understand it. I think this is the best idea I have heard so far.
> >>
> >> There is a subset of users -- we don't know how large -- who will never
> >> use yum because either (a) they won't have the bandwidth to use it, or 
> >> (b)
> >> they won't even understand that they need to use it.
> >>
> >> I still think that the Firefox technique will give us a better
> >> cross-section of data.
> >>
> >> Let me think about it.  I need to send a note to the docs folks anyway --
> >> part apology, part analysis.  I'll cc: this group as well.
> >>
> > 
> > Just pointing out that about half of the machines I install FC on
> > don't have X installed, much less firefox.
> 
> That's a good reason to put a tool in that does register new servers. 
> Well, maybe register is the wrong word.  That implies a lot of things to 
> me.  But certainly "get counted."  Which is different than "track."
> 
> I guess it's good to talk about the metrics we're interested in.  That 
> will drive how we build something:
> 
> 1. Tracking unique desktop users.  Also, how long do they use it for? 
> On the weekends?  On weekdays?  Do they stop after a number of days of 
> use?  One day never to return?  These are very useful statistics in 
> realizing what kind of job we're doing.
> 


that's more invasive than I think we should be.

we can count the total number of connections and we can count the
connections to which distro and on which days.

But we SHOULD NOT track a given ip to figure out what they're doing
and/or when.

That's inappropriate and I will resist it.

-sv





More information about the advisory-board mailing list