Re: The future of the release team

[Maciej Stachowiak <mjs noisehavoc org>]

On 22Sep2004 08:29PM (+0100), Mark McLoughlin wrote:
> > 
> > So basically I think the idea of splitting all these tasks among
> > different people is unlikely to work well.
> 
> 	That certainly sounds fairly reasonable, but in practise we've been
> splitting up these tasks very successfully. There is a small bit of
> overlap between the people doing the different tasks, but not much.

However I think there is a sense of shared responsibility for them,
even if they are being done by separate people.
 
> 	And I agree that you don't want random people taking up these tasks
> without the ability to do a good job. But its no different from any of
> the other responsibilities in GNOME - e.g. you don't find incapable
> people taking up the maintainership of unmaintained modules. 

Unmaintained module is a rare situation though. Usually you have the
same maintainer continue, or pass on maintainership if he
cannot. Random people volunteering without any kind of input from the
old maintainer is kind of a last resort.

I think these release tasks have a degree of community-wide
responsibility that requires a higher degree of accountability than
just "I volunteered randomly".

> Most people are aware of their capabilities and don't generally volunteer for tasks
> unless they're pretty sure they're not going to embarrass themselves.
> 
> 	What worries me more than the risk of things getting screwed up by new
> volunteers is the risk of people who would like to help might never get
> the chance to.

I think it's great to bring in new people, but it seems to me this
would require more coordination between the people doing the different
tasks, not less (as implied by the phrase "disbanding the release
team"), so the new people can learn the process. How about rotating
the membership a bit more? It used to be there was a different head of
the release team for every major release. Now it seems like Jeff has
been doing it forever.

The release team page itself says: "It's probably a good idea for
approximately half of the membership to change every 6 months to allow
many people to experience a full release cycle and learn about the
release process. This would also avoid stagnation." I wonder why this
approach doesn't create enough opportunities for new volunteers?

> 	It wouldn't always have been possible to do it this way, but things
> have evolved such that the release process chugs away with only
> relatively minor babysitting from us. And that's a good thing.

I am glad to hear things have improved so much. But, I also remember
the total mess that the process once was. I think the change is
largely due to having a strong set of people driving the release, and
I am not so confident things would keep running themselves if that
went away. It is like saying "development of module FOO is running so
smoothly, why even have a maintainer any more?"

That's really all I have to say on this topic though. I am not closely
involved in GNOME development these days. Mainly I wanted to pass
along my wisdom based on being on the front lines in the old days when
dates were set more in mud than in stone. I think it *is* possible for
a smooth process to get messy once again. 

Regards,
Maciej