Re: Opie's future

Am Donnerstag, den 17.11.2005, 06:07 +1000 schrieb Lorn Potter:
> On Thu, 17 Nov 2005 02:22 am, Michael 'Mickey' Lauer wrote:
> > Am Mittwoch, den 16.11.2005, 09:12 -0500 schrieb Neuman, Ben J., A&M
> >
> > IRM:
> > > I am a huge fan of the work you guys have done on Opie and am very sad to
> > > hear it's future might be in jeopardy. Is this true?
> >
> > It is - somewhat - true. If you recapitulate the history of Opie you
> > will notice that there was a time when we had ~20 contributors. Most of
> > them have jumped ship due to several reasons. The most important ones
> > are [in my subjective view of order of relevance]:
> >
> > * Sharp dropping the Zaurus from all but the japanese markets
> I don't agree with this one. Take the ipaq for example. It doesn't even ship 
> with linux, but that doesn't stop people using Opie and Gpe on them.

The ipaq world is much different from the Zaurus world. ipaqs ship with
WindowsCE - people who flash Linux to it do it because they already know
about the merits of Linux - among these people you don't attract that
much new developers. Take the huge amount of people into account who
submitted applications to the first Trolltech/Sharp application contest.
Among those were many developers who loved their SL5000d and later their
Sl5500 and who were so disappointed when Sharp didn't bring new models
that they quit their engagement into embedded Linux. There's not much to
discuss about this fact. Look at killefiz. Look into OE. Leaving Opie
aside, 90% of applications for the Qtopia/Opie platform are orphaned.

> > * Opie still using the (outdated) Qt/Embedded 2.3.x
> I don't agree with this one. Qt/E 2.3 is still perfectly usable.

Usable, yes. Comfortable, no - especially when better tools are out
(Qt3,Qt4). This is open source, remember? Open source developers
sometimes have more demands than your professional customers.

> > * OpenZaurus being binary incompatible with Qtopia apps
> This is the major factor, I believe. Also add no easily available cross 
> compiler that is compatible with OpenZaurus softfloat. Development is too 
> high of hurdle. No one wants to fuss around with openembedded when all they 
> want is to develop applications.

Yes.

> > * Unclear Opie roadmap, goals, and identity
> I do not think Opie _ever_ had these, except identity, which I do think Opie 
> has it's own. and I do not recall ever seeing any formal discussions about 
> these.

Exactly - and this is why we are doomed. We had (semi-)organized chaos
and the most we got out of it was people who put stuff into Opie cvs and
then vanished and left us to mainiain more and more half-baken
applications. Some of us wanted to leave Opie as a Qtopia-clone, some of
us wanted to bring Opie to a new level where it could have stood out as
a seperate and attractive entity. There was no agreement. New developers
are not attracted by chaos, new developers are attracted by solid
platforms, richs APIs, and projects with clear goals and concise
roadmaps.

> > These reasons not only affected Opie directly but also third party
> > authors writing applications based on the Qtopia API.
> >
> > Nowadays, a major part of the remaining Opie contributors seem to be
> > more interested in going away from Qt/Embedded since 
> > a) X no longer is bloaty or slow compared to Qt/Embedded and 
> I'd like to see clear evidence of this. In my experience, Gpe is too slow to 
> use on a 5000d.

I can't confirm that. Let's bring numbers and benchmarks.

> > b) a small community like us 
> > no longer can afford ruling out millions of applications just because of
> > direct framebuffer access.
> I don't think the majority of linux desktop applications are suitable for use 
> on a Pda. You still have to touch them up for small screens, and 
> keypad/button navigation.

Of course. But as we have already seen in the short life of maemo, this
is done much more quickly than to port an application over to another
toolkit plus application framework.

> We need to concentrate on developing/maintaining just a few good applications.

That's why we should've made a roadmap many months ago -- when we had
more people working on Opie we could've focused. Now it's too late.

> There are limitations in any software, but we do have x11 access with Qtopia 
> and Opie - xqt (which should probably be integrated into Opie in an optional 
> way).

Xqt is a nice proof of concept, but it's sluggish and needs a lot of
memory. For people who are satisfied with Opie and just need to use one
or two X apps once in a time, it may be ok. If you use many X apps
regularly, why use the additional memory overhead and indirection?

> > That aside, I'm still very interested in writing open source software
> > for mobile linux devices and I hope that -- after completing my Ph.D.
> > next summer -- I can return to writing software for X11-based mobile
> > devices using Qt/X11 4.x and/or Gtkmm/Hildon.
> 
> 
> > The future of Linux on mobile devices has just begun and it's exciting
> > to be a part of it.
> I do not see Nokia's experiment as the start of some great new Linux embedded 
> Utopia.

Tss... You turned what was meant as a general statement (precisely
indicated by three carriage returns) into a forward pass to bash Nokia.

> The future seems to be smartphones (qtopia phone has sold way better than 
> qtopia pda). Pda sales and interest is dwindling.

So what's your point with that statement? Should we all give up on PDA
programming then? Besides, if I remember correctly I was talking about
"mobile devices", not just PDAs.

-- 
Regards,

Mickey.
------------------------------------------------------------------
Dipl.-Inf. Michael 'Mickey' Lauer <[email protected]>
------------------------------------------------------------------


_______________________________________________

http://opie.handhelds.org/cgi-bin/moin.cgi/DeveloperWikiIndex

Opie-devel mailing list
[email protected]
https://handhelds.org/mailman/listinfo/opie-devel