[maemo-developers] Maemo & Linux mainstream (was Re: Projects Nokia should support (yours?))

[From: Quim Gil]

Thanks for the very good feedback!

There is a common thread going down to the Maemo platform and its
relationship with Debian/Ubuntu aka Linux mainstream. This is a
discussion started many times, let's hope this time we move it forward
to higher stages.

Answering to Till and Neil, who came up with requirements difficult to
combine: mobile uniqueness versus API compatibility across several
platforms.

ext Till Harbaum / Lists wrote:
> Is this really clever? Is this really what Nokia has to say about their new 
> plattform? "Look, our new plattform is finally good enough to run
> software that is well established on these other plattforms"?

50% agreed. It would be deceiving that all the Fremantle starts would be
ports. But it would be <your adjective here> to ignore all the good work
done in the Linux desktop and not bring anything from that to Maemo 5
users (most of them probably never Linux users before, by the way). We
need to find a good balance reusing what exists in order to concentrate
in whatever doesn't exist yet.

Agreed because indeed projects like e.g. Gimp, OpenOffice.org and etc
ported to Maemo are interesting from a technical and experimental point
of view, but as such are not the cases that would drive the platform
innovation and the products its end users will dream about.

The problem is not that much on the performance side. Performance is of
course a problem but in our opinion not as big as the differences in use
cases and UI requirements for a small touchscreen. Gnumeric in your
pocket? Sure, but...

On the other hand, Nokia sees a value in the fact that standard Linux
components do run in your pocket without needing embedded versions.
Maemo has diffs with Debian/Ubuntu we want to minimize, but still Maemo
is mostly a Linux standard OS. Just compare with the average Linux
embedded variants shippedm in devices fitting in your pockets.

In this sense there is nothing intrinsically bad in desktop applications
keeping the very same engines and concentrating on a UI focused on the
mobile use cases. For instance, nobody is thinking of creating a new
browser engine since the way to go is optimize the current desktop
browser engines (and the desktop users benefit from this too). But note
that nobody is seriously attempting the keep a desktop browser UI in a
mobile platform.


ext Neil Jerram wrote:
> I sense two fundamental problems here.
>
> 1. You (Quim) are looking for applications that will help
> differentiate the next Internet Tablet.  But differentiation is
> contrary to the interests of the majority of the community.

I don't know developers not willing to produce software that users find
amazing and fall in love with. Ok, in fact I know some  ;)   but you get
my point.

Differentiation, in the way Nokia understands differentiation, is
nothing we are looking for in this call. What Nokia sees in Maemo 5 is a
very good chance to get pure community (and yes, open source) software
in the hands of pure end users - and impress them, for the software
itself and they way it is being developed. Just like many of us were
impressed when we discovered community software development in the
desktop, now in the mobile context.

If the Fremantle stars become stars in other platforms as well, all the
better!



> 2. The timescale is too short.  Either people already have most of
> their time committed for the next few months, or they don't.  I would
> venture that those with time on their hands are mostly unlikely to be
> people who could pull a project together in this timescale, and to the
> quality that you are looking for.

Short time and shortage of resources for impressive native applications
developed from scratch on a volunteering basis? Probably, yes.

But we think time is not too short for

- Push existing projects to a certain goal.

- Merge several one-guy adventures into something stronger.

- Develop brilliant ideas connecting existing components and new features.

- Develop from scratch brilliant ideas requiring a light execution via
e.g. Mozilla add-ons, Python etc.

We could find many examples already in the Linux and Mozilla
communities. These days we are seeing good examples coming also from the
emerging Android and iPhone platforms. Something to learn from there?


> (2) is self-explanatory.  To expand on (1) a bit...  My time for free
> software is limited, and the best outcome for me is if I can use that
> time to write something once, and then have it automagically show up
> and run on the widest possible set of devices: my computer, internet
> tablet, phone, and so on.

Yes, we understand. There is no perfect solution for that today. We see
different approaches to tackle that going through Mozilla add-ons,
Python runtime, Qt API and probably more.

> (Note that "writing" here includes activities like packaging.  In an
> ideal world, I would only have to package each new version of my
> software once, upload the source package to Debian, and everything
> else would follow from that.  Note that this has now been achieved for
> the Openmoko phone, so it is entirely possible.)

Even if it's technically possible, there is something no technology
solves (at least nowadays): the use cases for someone in the move with a
device in their pockets are different from someone sitting with a
computer on the table. For many apps just this is critical.

And one idea: porting a mobile application to the big desktop is
actually simpler than the other way around.

> So, for me, anything that requires me to "port", or to do something
> differently for the Nokia tablets, is a negative.  A couple of
> specific examples are
>
> - the /var/lib/install thing in the first 770 OS ... thankfully now
long gone
>
> - the existence of hildon.

Hildon is available in Debian and Ubuntu. It covers features important
in the mobile context that GTK+ is not covering as for today. If you
really dislike it you can circumvent it in Maemo.

> And so here are a couple of project ideas.
>
> 1. Work on a Gtk+ theme, and changes if needed to the upstream Gtk+
> project, so that any existing Gtk+ app will run (without any porting)
> on the tablet and look as good as hildonized apps do today.

Interesting idea. Maemo is already investing a lot improving GTK+ in the
direction and I'm sure the GTK+ maintainers will love to see more
contributions done in that front.

> 2. Organize the system so that the Debian arm repository can be used
as is.

Very interesting topic I'll leave for a separate email.

> If you did that, there would be 100s of existing apps that would run
> and look good - which for me would be a far more powerful selling
> point than one or two specific applications written specifically for
> the tablet.

Yes, and no. Having hundreds of applications available is indeed a good
selling point worth pursuing. But nothing conceived for a PC and the
related use cases will squeeze the possibilities a mobile device can bring.

Theater was essential for cinema, but the great films were different
from theater plays with a camera in front. Traditional media were
essential for the WWW but great web projects were different than online
versions of newspapers and stuff. We are just in the same situation with
the jump from the PC to the mobile. Like Tim pointed out above.

-- 
Quim Gil
marketing manager, open source
Maemo Software @ Nokia