[maemo-developers] Public maemo repository

[From: Marius Vollmer]

"ext Kees Jongenburger" <kees.jongenburger at gmail.com> writes:

> from what I understand it' maemo that has been assembled  from well
> known wood sources (debian/linux/glib/gtk). I just can't imagine that
> you could have been that creative/productive with hildon if you also
> had to manage external deps/communication.

Why not?  Hildon was one of the first things that went fully public
after the 770 launch.  We had to deal with the consequences right
away.  We might not have been perfect in doing that (enormous, sudden
API breaks, no proper releases, etc).

We didn't provide Hildon in the context of a continuously evolving
distribution, but doing that would not have slowed us down, I think.
We just would have had to learn some lessons earlier.

> Nokia does not make it a secret that they choose for "strong
> upstream projects" and the last moves (the ubuntu mobile stuff)
> shows that apparently something was created that will be usable on
> different platforms. so you are really going the distro/generic
> aproach where the person who packages a package and the person who
> created software are not the same.

Yes, I don't insist that Nokia has to run the "maemo Distribution".
Ubuntu Mobile could make the maemo distribution unnecessary.  In fact,
my first reaction when I heard people start talking about a maemo
distribution was: Isn't it a bit late now?  Can't we just wait for
Ubuntu Mobile?

> Anyway i wonder what part of maemo is to be attractive if the hildon
> development goes "outdoors".

The community will still cluster around maemo, of course: the planet,
garage, etc, and maybe the distribution.

> Could you elaborate a litte on the choice for debian as opposed to
> openembedded or  the  openwrt kind of distro's?(small fast ships that
> allow water-skying :P )  is that the strong upstream projects thing?

I can't talk about the initial choice as it was already made when I
joined Nokia (and it was in fact one of the points that convinced me
that Nokia got it right before the 770 was launched).

Later, when enabling real package management in IT OS 2006, I talked
some with the people involved in OpenEmbedded, Emdebian and looked at
ipkg, etc.  At that point, we already had the SDK based on Scratchbox
and the Debian packaging tools were used internally and externally.
Neither of the alternatives seemed to offer enough advantages to
justify a switch.

The 770 and N800 hardware has no problem running dpkg and apt,
Scratchbox is (on average :-) a nice cross-compiling substitute and if
anything Operating Systems for mobile devices will converge with OS's
for general purpose computers (that's why Nokia went with Linux, X11,
Gnome, etc in the first place).  Thus, there is no point to go with
some specialized, 'niche' approach to save 2 MiBs of flash[1].

Projecting this into the future means that we should work to make a
existing mainstream OS distribution useable as the base of the
Internet Tablet OS.  That could be Debian GNU/Linux or Ubuntu.  Our
vehicle to get there could be our own maemo distribution, or we could
skip that step.

[1] Yes, I admit to freely waste hardware resources to save human
    resources and to stay in the mainstream.  Old time embedded
    engineers might get gray hairs about that attitude, but then they
    should go and program washing machines... ;-)