Re: What would you do to encourage application developers on GNOME Mobile?

[Murray Cumming <murrayc murrayc com>]

On Thu, 2010-04-08 at 08:56 +0200, quim gil nokia com wrote:
> Hi Murray,

Sorry for not replying until now. I got distracted.

> In the context of this discussion it is useful to that Maemo 6 API = MeeGo API.
> 
> > If we are talking about Maemo 6, then you need to know that it will be a
> > large amount of work to make GTK+ and libhildon properly supported on
> > Maemo 6.
> 
> Indeed, if you plan to map all the DUI libraries sitting on top of Qt. What about mapping the Qt UI libraries instead?

That would probably be easier, yes, but it's presumably a matter of
using Qt in a certain way. I don't believe in the just-recompile mantra.
It's not clear yet though.

> > So far Nokia has only said that the community may support it.
> > But:
> > a) That won't be enough, just as it wasn't enough to get Qt supported on
> > Maemo 5.
> 
> Manpower was the only problem, right?

Maybe, though there was also a lack of adult supervision and a refusal
to accept that 
a) API additions were necessary.
b) It was _not_ easy to just reimplement custom widgets in applications
so they'd look exactly like what Hildon offered, so the development
direction was misguided.

I still don't think the official port is really good enough, though it
makes Qt on Maemo 5 really viable. But it was quite awful before. 

> > b) The community won't have access to the necessary information until
> > far too late. This needs people who have are on the inside at Nokia.
> > (like, say, Openismus. Ahem.).
> 
> MeeGo plans to have a first release in May with all the Handset UX whistles. I would even say that you could go to qt.gitorious.org and start the work now. Maybe there are relevant bits still available inside Nokia only (honestly I don't know, you might know better than me) but most of the DUI code is out already and Qt itself is already integrated in the MeeGo repositories since Day 1.

I don't think anything is very clear. Maybe it will be then, though I
expect to have the usual complaints about lack of clear API advice.
However, we'll be pushing for clarity, partly by our David King working
on developer documentation/assistance for maemo.org.

> > c) It will probably require changes to the GTK+ and hildon API on Maemo
> > 6, just as Qt has API changes on Maemo 5. So it will get conceptual,
> > which will slow things donw.
> 
> It all depends on the approach you want to take on this. You could choose a conservative approach making the current Hildon work on top of MeeGo's GTK+. This probably means that Hildon apps in MeeGo look and behave like Maemo 5 apps, in exchange of less work with bindings and less code changes for app developers.

That's obviously far easier, and probably a necessary first step, in
case the rest never gets done.

It doesn't seem very useful in the long term though. There was no point
in having GTK-only (not using Hildon) or QT 4.5 (without massive hacks)
apps in Maemo 5, and there would be no point in having GTK+/Hildon apps
in Maemo 6 that feel equally out of place. 

> As a stepping stone looks good enough, and then you can decide on further steps based on feedback and interest.

Yeah, I think this should be done. It's an easy decision even if it's
only a small part of the plan, and things would be very bad if not even
this is done.


-- 
murrayc murrayc com
www.murrayc.com
www.openismus.com