[Sugar-devel] [IAEP] SLOBs Position on SoaS

[Peter Robinson]

2009/9/16 Philippe Clérié <philippe at gcal.net>:
>> Isn't there a wider question first? the one that asks if Sugar Labs is
>> actually interested in being a distributor rather than just an
>> upstream.
>>
>
> Sugar Labs needs to be a distributor because:

I disagree.

> 1) You need a product to market. The comparison with Gnome does not hold.
> There have always been distributions that made Gnome their official desktop
> environment, even very early on. That is not the case for Sugar. Whether in
> Fedora, Debian or Ubuntu, Sugar will always be a secondary desktop at best.

That is in fact completely incorrect. KDE is not the primary desktop
in Fedora but there is a team that makes it a first class Desktop in
Fedora. The team in sugar that deal with Sugar actually do most of
their work in Fedora and then basically add a few minor things to make
it SoaS. No reason they couldn't do all of that upstream, it also
means that Sugar is definitely not a second class desktop in Fedora.
Its basically the same class as that of SoaS.

> 2) Sugar needs a dedicated distribution. Geeks will always be able to
> navigate distributions installation procedures to choose Sugar as a desktop.
> I am afraid that is not the case for the vast majority of teachers out
> there. They need something that they can pick up and run with.

There's nothing to stop that from happening in another distribution.
You could easily have "Fedora Sugar Desktop" and have it basically
exactly the same stability and ability to run off a USB stick etc.
Also in my experience of doing a pilot in the UK a lot of teachers are
confused by "Sugar on a Stick" and you basically get blank looks and
they have no idea how to boot their laptop or any other device off a
USB stick. They see them as a way of moving documents around, not
something to run a OS from.

> 3) A distribution can be a source of revenue. Inevitably, as the developers
> of Sugar, you will be asked to provides services in one form or another.
> You'll want a distribution you control for efficiently doing that.

It can be, but is there plans to?

> 4) Maintaining a distribution is an invaluable experience. You'll hit
> integration bugs before they are discovered downstream and those reported by
> downstreams will ring a bell and be easier to find and fix.

And in the reverse of that argument by using an upstream such as
Fedora you don't need to learn about all those issues or you can use
all the learning that Fedora or another distro has gained by
supporting millions of existing users and hence you don't need to
re-invent the wheel hence saving a lot of engineering time.

> 5) A distribution will give you control. Other integrators won't necessarily
> follow your philosophy in how Sugar is packaged and/or deployed.

That's certainly not the case in Fedora. The Fedora and Sugar
philosophy in most cases is very similar.

> There are probably other good reasons that people can come up with.
>
> The last thing I'd like to add is that it's not an either/or proposition.
> Sugar Labs needs to be *both* upstream and a distributor.

That depends on which way you look at it actually. There's advantages
to both approaches.

Peter