debathena-cluster (and -workstation) should enable popcon

[broder]

We should enable  popcon on debathena-cluster and -managed (or -workstation - whatever) machines.

This would largely be useful for showing that some of the packages that we've upstreamed ourselves are actually being used on 1000 machines or so. For example, config-package-dev currently only has 63 votes in Ubuntu.

[jdreed]

I understand the desire to do this (and agree with it in principal), but adding a package that phones home outside of MIT is a major change and we should run it by release-team first, especially for the MIT-owned cluster machines. And it's not at all clear to me that some -workstation/-managed users won't be completely freaked out by this. I wonder if there's any good way to have this be opt-in for non-cluster machines, though the downside there is that most people are likely to click no, the same way they click "No" for the "Should this crash report be sent to Microsoft/Apple??" dialogs in Windows and OS X respectively.

[broder]

The conclusion of release-team seemed to be that we shouldn't turn this on on workstation and cluster machines, and that regardless, popcon only tracks packages that have been accessed, so we'd be hard-pressed to use it to inflate our own ratings anyway.

[price]

[price:]

This is a surprise. What was the reasoning here? If a package is
being used on a workstation or cluster machine, it really is being
used -- nothing is inflated.

[jdreed:]
Hrm, perhaps I misunderstood. I didn't think we definitely weren't doing
it, I thought we were tabling discussion of any metrics collection until
later (June) once we know exactly what data people are looking for and how
we can collect it.

[tabbott]

It is also worth noting that even if MIT doesn't find popcon's information as useful metrics, enabling it on cluster machines may still be useful as popcon is definitely a metric used by Ubuntu and Debian upstream.