David Powell wrote On 10/26/05 17:10,:
> On Wed, Oct 26, 2005 at 04:10:15PM -0600, Bonnie Corwin wrote:
>>Depending on how things go (which is hard to project at this point), it
>>might not be an issue. If we get a manageable number of third-tier
>>sponsors who do a good job, we shouldn't have a problem.
> How exactly do you determine if they are doing a good job? Again,
> this is not a question of whether the putback is successful or not.
> It is a question of whether the tier-3 three engineer is adequately
> and accurately communicating Solaris engineering practices to the
> contributor, not to mention appropriately representing Sun.
> The only system I could see possibly addressing this issue is one
> where every mail the tier-3 sponsor sends to the contributor, and
> vice versa, also CCs the partner engineer. Mistakes can be corrected
> as they occur. This has many problems:
> People are easily dropped off of CC lists in 1-1 conversations. I
> would expect this to happen frequently with conversations initiated
> by the contributor, since the contributor doesn't know "the other
> guy" and has no reason to remember to include him.
> It looks really bad when we correct each other, worse if
> disagreement breaks out. The sponsor should be authoritative.
> Detecting/dealing with errors of omission is tricky. At what point
> does the partner stop waiting for the sponsor to mention something,
> step in, and say it themselves?
> Lastly and most importantly, the contributors are doing important
> work and shouldn't be treated like guinea pigs.
>>Does that sound reasonable?
> Not yet.
Then what do you suggest?
This won't be perfect. Perfect would be our most senior engineers
having the cycles/interest to cover this work completely.
I believe engineers interested in this work will be interested in
building relationships with external developers. And therefore will do
their best to represent Sun and explain engineering practices.
With this proposal, we will probably know about problems when putbacks
get backed out or partners make us aware that someone isn't working out
or an external developer contacts us with complaints.
In addition, we can watch data such as:
- # of third-tier sponsors being added
- how quickly requests get picked up
- how long putback cycles take
- whether it's hard for third-tier sponsors to find partners
- whatever else people think would be worth tracking
And we can set a time period (3 months?) and assess the data and the
feedback to see if we should continue or regroup again.
Not perfect, but I think a path worth trying.