Hi!

(I have some pressing private matters, so I will probably be pretty
absent this week, just so you know)

Anyway:

Andreas Raab <andreas.raab at gmx.de> wrote:
> Hi -
> 
> >>I kinda agree and for me it triggered the interesting question 
> >>what nowadays process is.
> > 
> > At the moment there is no process. What we do have is the Janitors team
> > which has been looking into harvesting related issues, granted that Team
> > has not yet taken on the big question about exactly how the process
> > could work.
> 
> Err .... I am perplexed. If I look at 3.8 I find quite a number of fixes 
> and enhancements. Surely, there is some process in place right now, or 
> else how did these changes get in?

What I meant is that there is no "harvesting process" in place, but
indeed there is a bit of harvesting going on - mainly Steph and Marcus I
think.

And... Michael and others harvested for 3.8 as part of the release
responsibility.

> >>With the guides being fired and all I wonder 
> >>who is in charge to actually accept contributions?
> > 
> > Hehe, that is a misleading sentence. First of all, the Guides didn't do
> > much harvesting at all and the Harvesters as a group was not equal to
> > the guides. Secondly, the issue of stalled harvesting predates the
> > forming of the Coordinator group.
> 
> Well, I did see the harvesters being part of the guides setup.

Ehm, (thinking hard)... no, I don't think those two groups were related
in any way actually, except for overlap of course.

> At least 
> I don't see anyone calling himself a harvester anymore, nor do I see 
> anyone referring to harvesters any longer.

Right, but that fell apart before Coordinators. In fact, it was one of
the driving forces for "doing something".

> Do I understand correctly 
> that for the current setup the "janitors team" plays the role of the 
> harvesters, or is this incorrect?

No, but they have the role of figuring out how to handle the immediate
situation, which includes dealing with BFAV<->Mantis and thinking ahead
about the Harvesting process itself.
 
> > But the important question of course remains, what model/process do we
> > want to have?
> > 
> > - I know the Janitors team has been doing some thinking. I would like to
> > hear more from them in this, because that is after all the team formed
> > in that specific area. Ken?
> 
> It would be interesting to hear, yes.

Yes. But I also want to say that I am not trying to make it sound like
this is the Janitor's job - I just am aware of them having thinked and
discussed it quite a bit - and especially the related stuff like Mantis
etc.

> > - The packages Team is meant to produce a partitioning during 3.9 which
> > means we get people assigned to look after portions of the image as
> > packages. This will IMHO greatly improve the situation since all FIXes
> > should be handled in a distributed fashion. Unfortunately it seems that
> > Team is a bit stalled too, at least when it comes to that specific task.
> > Avi?
> 
> This is basically TFNR under a different name, right? Why will it work 
> better this time compared to when you tried it? If I read the current 
> stall correctly, it's the same problem that TFNR had - lack of 
> maintainers. Any ideas how to address it differently this time?

Yes, this is the same thing - and yes, I have ideas. But I am not the
team leader. I actually don't think the lack of maintainers is the main
issue at the moment (but the lack is of course a problem, but I think we
could find "enough" to get it going) but the problem is getting out of
the gate with an initial partitioning - no matter how "faulty" it is.

We got stuck at the exact same place last time - and we had a bunch of
people lined up - but we ran out of gas trying to whip up the first
categorization/PI-ification. That is why I have repeatedly said - don't
get stuck there - just whip something up fast and quick.

But I know "fast and quick" is work too. :) And no, I am not blaiming
anyone in particular here - especially not Avi - we all have lack of
time, and he has also signalled that.

> > - The release team leader should IMHO focus on dealing with ENHs and
> > other planned additions for a release. That doesn't mean the leader does
> > all the work, but it means that the Team formed for a release takes the
> > overall decisions and actions regarding ENHs and additions, of course in
> > synch with the package maintainers mentioned above because they are the
> > ones in charge of the various packages.
> 
> Correct me if I'm wrong but with "TFNR rules" isn't the *maintainer* 
> responsible for integrating enhancements? What has the release team to 
> do with that?

Well, it was a bit of "loud thinking". :) What I was trying to say is
that yes - the maintainers are responsible for everything really, but
some balance would be needed or otherwise the release team would be
kinda... well, they need to have some saying too I think.

This is uncharted waters - but I do think having both (maintainers and a
release team) working in tandem somehow is what we want.

> > Now... the above describes a model/process which really doesn't have a
> > dedicated group of harvesters but instead relies on the image being
> > partitioned into packages with maintainers and having dedicated release
> > teams, and I think that is the direction we should move.
> > 
> > What do you all say?
> 
> Same thing as last time: Sounds nice if it works ... but ... where are 
> you going to find the maintainers? (and by maintainers I don't mean name 
> tags I mean people actually putting effort into it)

I agree. I do think we have at least enough to get it started and having
it grow over time.
But a lot of pieces of the current image will probably end without
maintainers and thus in a bucket which we will have to gather a group to
deal with.

> And besides that, given that none of this is in place what happens in 
> the meantime? Nothing?

Eh, well, no. Doug is the release team leader for 3.9 and my impression
is that he is loosing up the model so that we can get more immediate
access to the streams and in that way "get going in the meantime". Doug?

> Cheers,
>    - Andreas

cheers, Göran