gmane.linux.laptop.olpc.sugar

David Farning | 14 Aug 20:31

[sugar] Release Cycle

Now that Sugar Labs has released .8.2 to OLPC it is time to revisit the
release cycle issue.

I have the feeling that most of us agree _in_principle_ to the idea that
an established release cycle is  important.

As part of his Ph.D. Martin Michlmayr has done some interesting research
on the topic.  He provides a good introduction to the topic in his talk
'Open Source Speaker Series: Release Management in Large Free Software
Projects'.  [1]  His thesis is also available. [2]

Some thoughts on on how the release process applies to Sugar Labs.

- Background -

1.  Software engineering is hard.  There is no silver bullet.  No design
methodology, governance process, or optimistic belief is going to ensure
the success of Sugar.

2.  Sugar is Open Source.  

3.  Sugar Labs is a community.  At the end of the day, it does not
matter why Sugar Labs chose the community development process.  It is
here and we are stuck with it;) both good points and bad points.

- Stakeholders -

4.  Are our stakeholder happy?  The primary goal of any software project
is to ensure that its stakeholders are happy.  This happy is not a '70
kind of happy.  Rather, it is a happy where our users continue to chose

(Continue reading)

headers

Marco Pesenti Gritti | 14 Aug 22:39

[sugar] Release Cycle

David Farning wrote:
> Now, it is time for our stakeholder representative to sit down and hash
> out a plan for the up coming release cycle.  I am not naive enough to
> think that we can immediately forget our past grudges or politics.  But,
> I am optimistic enough to think that we can agree enough to set mutual
> goals and allocate resources to those goal...for one release cycle.  The
> best thing about release cycles is that in six months we can revisit our
> goals and asset reallocation. 
>   

Hi David,

I don't think it's realistic to plan for an all hands meeting on the 
short time. Most of the OLPC developers and management is busy getting 
the release out of the door.

* As far as I know Red Hat doesn't currently have *direct* plans on Sugar.
* Greg has been doing some great work on starting to hash out 9.1.0 
plans for OLPC. We are pretty well covered there and I'm confident 
requirements there will get increasingly clear.
* SugarLabs concrete needs are somewhat nebulous to me. Can we  fix 
that? We will have to rely on community help, since we can't currently 
fund work.
* There is a lot of work, like stability and performance, which will 
obviously benefit everyone.

My suggestion it to use this list and the SugarLabs wiki to plan the 
next release cycle. This will allow OLPC, SugarLabs and the whole 
community around Sugar to participate.

(Continue reading)

headers

David Farning | 15 Aug 05:41

[sugar] Release Cycle - Period

The first step in establishing a release cycle will be setting the
period.

Here again, I think that we all agree in principle on a sixth month
cycle.

>From Sugar Labs perspective the actual time period is quite arbitrary.
The lower bound is how quickly we can effectively iterate through the
innovate-stabilize-release cycle.  The upper limit is how long we can go
before a release becomes stale.

>From a Redhat perspective something that lines up with the Fedora
release cycle makes sense.

>From a OLPC perspective the issue is more complicated.  The cost benefit
analysis boils down to the cost of pushing a new release vs. the benefit
gained by improvements in the new release minus the cost of supporting
each additional release.  I will happily leave the math on that on to
Jim;)

With these facts in mind, Sugar Labs doing six month releases seems
optimal.  OLPC can pick up releases bases on their needs.

thanks
dfarning

headers

Marco Pesenti Gritti | 31 Aug 01:38

[sugar] Release Cycle - Period

On Fri, Aug 15, 2008 at 5:44 AM, David Farning <dfarning at sugarlabs.org> wrote:
> The first step in establishing a release cycle will be setting the
> period.
>
> Here again, I think that we all agree in principle on a sixth month
> cycle.

Yeah we discussed it in the Sugar meeting a couple of weeks ago and
there was agreement.

I updated the tentative schedule:

http://sugarlabs.org/go/ReleaseTeam/Roadmap/Tentative_0.84

> >From Sugar Labs perspective the actual time period is quite arbitrary.
> The lower bound is how quickly we can effectively iterate through the
> innovate-stabilize-release cycle.  The upper limit is how long we can go
> before a release becomes stale.

Six months is probably a bit too long from this point of view. Though
OLPC is planning 8.2.1 in December, which means we will maintain the
0.82 branch alive for a long time.

> >From a Redhat perspective something that lines up with the Fedora
> release cycle makes sense.

We would be 2.5 months in advance of the Fedora cycle. I think that's
pretty good.

> >From a OLPC perspective the issue is more complicated.  The cost benefit

(Continue reading)