[maemo-developers] Public maemo distro

From: Marius Vollmer marius.vollmer at nokia.com
Date: Mon Jul 23 14:23:31 EEST 2007
> Alright, so you want me to play evil's advocate role.

Heh, don't put yourself on the evil side, that's my spot! :-)  

> > Change means work and there needs to be a reason for it.  What 
> > I find to be usually missing is however a good justification 
> > of the status quo.
> 
> The status quo works. The current process accomplishes its basic
> mission.

That's the _minimum_ requirement for something to be acceptable as the
status quo, of course.  I hope that is not the best justification we
can find.

> There are things to be improved but a public maemo distro is not the
> only or automatic answer.

There are always things to improve and there is always more than one
way to do it, thus this line of argument hasn't much weight either.

One nice alternative to getting serious about the maemo distribution
ourselves is to let people do it that know more about it, such as
Ubuntu.  If you have other alternatives in mind, please share! :)

> > I would like to see the crystal clear reasons why Nokia decided to
> > [derive from Debian sources, including the packaging, but chose to
> > build something out of it that is very much unlike Debian.]
>
> Who is in a better position to know than you?

Ohh, many. :)

Here is the history of my small corner: I joined Nokia shortly before
the 770 launch and inherited the anemic "Application Installer"
control panel applet that some might remember from the IT OS 2005
days.  It ran dpkg as the "install" user chrooted to /var/lib/install.
Packages made for this installer could not declare dependencies on the
packages that made up the OS.  Apt was not part of IT OS 2005.  I
guess this was the first WTF moment for people who had heard that the
new Nokia Internet Tablet is "based on Debian".

Thus, the original design of the IT OS was far away from a typical
GNU/Linux distribution.

We have come a long way since then, a bit because I assumed that the
original limitations of IT OS 2005 were born out of conservatism (with
respect to the approaches of Symbian and Windows), carefulness and
time limitations.  So we went ahead and 'fixed' the Application
Manager for the next release to use apt and put 3rd party packages on
equal footing with operating system packages.  No crystal clear,
tenable reasons in favor of the crippled Application Installer
approach surfaced during this redesign process.

> > This work is mainly caused by missing the boat in the first place
> 
> If the boat was the 770 launch deadline, the boat wasn't missed.

That's another boat. :) The missed boat that I am talking about is
that we should have put out the open parts of the IT OS 2006 release
(the one with the 'fixed' Application Manager) in a repository,
provide a matching "extras" component, and declared the two to be the
maemo distribution.

I say this with hindsight, of course, and I don't try to point fingers
at anybody.  I was there and I missed the boat more than most.

We did something that was quite close, tho: the maemo 2.x SDK
repositories nearly contained the open packages of IT OS 2006.
However, some packages were modified for the benefit of the SDK and
this made them break on the device.  This was considered unfortunate
but OK since the repository was only for the SDK.  I guess this was a
second, smaller, WTF moment for some people.

> > a distribution is a good tool to reduce gratuitous work.
> 
> Yes, sure. I see you are concentrating in the code management,
> perhaps overlooking the energy that takes to satisfy in a *public*
> distro human expectations, communication, contributions.

Well, I see getting serious about the maemo distribution as one
ingredient in _addressing_ ours and others needs, expectations, etc.
I don't think new expectations will arise, but old ones might be
voiced more clearly and the community might grow larger since 'Nokia
starts to get it', thus bringing more voices to the choir.

> It is not difficult to have a public distro with users and
> contributors unhappy because the human side is not working.

Again, the unhappyness is not caused by having the distribution, the
unhappyness exists and is addressed by improving the distribution.

> In fact just a few distros (from many) succeed in their mission.

But they didn't fail because they are a distribution and they would
not have succeeded if they would have been something else than a
distribution.

> Failure factors are based more on human factors and context rather
> than efficiency of the code alone. Just make memory and look around.

Now we are talking about maemo as a whole.  Will maemo succeed?  What
is the mission?  What are the competitors?

The scope of that discussion is indeed much bigger than what I try to
talk about here.

More information about the maemo-developers mailing list