[Development] The place of QML

Fri May 18 20:05:16 CEST 2012

Hello,

I've been following this thread since it started. My impression is that Qt
people believes we (users-devs) want to force you into further developing a
C++ toolkit (which might or might not be outdated). But I understand that
you will do work on what are *ordered* to work on, by your boss. It's
understandable.

Others think, that this discussion is about Qt Devs & team forcing QML (a
still immature & incomplete toolkit) into the community. While Qt Widgets
have been serving this community very well since years ago, these people
think this is a step backwards. QML *might* have a great future but it's
not even complete yet. Now it seems it's a fight between QPainter and QSG
(it think that's irrelevant).

Qt Devs have made pretty clear that QML will be a first class citizen along
C++, but in reality, there is a sense about Qt C++ being abandoned.
Specially when QML is still difficult to use from C++ and no work is
planned to be done (unless the community wants to step in) and even those
modules are marked as "done" (no further work by Qt Devs).

My point is a bit different, it's about the apparent disregard for Desktop
Components or 'legacy' widgets, I think it will contribute a little. QML
seems to be "the future" or at least that's how you want to market it.
Good, i like it myself. But even when "the future" now seems to be plagued
of futuristic "natural" interfaces, animated tiles and what not (I've
working hard on this as my PhD is about Human-Computer Interfaces (HCI)),
there are several strong critiques about it, by very renowned people, see
Donald Norman's: "Natural interfaces are not natural" [1]. Yes, he mainly
talks about gestures & touch, but that's the central point in creating
interfaces with QML.  Years, decades, --no, centuries of man-hours-- have
been poured into the current interfaces, they work just fine because of
this. Incomplete, delayed or ignored support for so called 'legacy' widgets
is a heavy decision, is it based on reliable/exhaustive research? Other
have been studying Apple interfaces and have found they are nice, but, in
fact, difficult to use for many people, specially those >30yrs. People are
dropping well researched interfaces and replacing them with MS Bob[2] style
interfaces.

MS Bob/Touch friendly interfaces will not replace traditional interfaces.
So QML/Silverlight/Cocoa's flashy UI's will not fully replace traditional
apps. At the end, millions of real apps for real work will safely rest in
well researched UI grounds. So please, i love QML, but think twice before
really ignoring legacy QWidgets, make it easy to use current C++ API's with
the new QML elements (QML's base items are not even available to subclass),
pour a little more work on Desktop Components or, if not, at least create
*real* tools for us to develop UI's. No, you can't develop a serious UI
just in Creator, and the PS exporter script is really far from perfect,
even Blackberry has a better exporter with it's Cascades QML Exporter [3],
maybe you could coop, why 2 exporters?.

Ariel

[1] http://jnd.org/dn.mss/natural_user_interfaces_are_not_natural.html
[2] http://youtu.be/5teG6ou8mWU?t=2m54s
[3] http://devblog.blackberry.com/2012/05/cascades-builder/

On Fri, May 18, 2012 at 5:40 AM, Иван Комиссаров <abbapoh at gmail.com> wrote:

> You need users who will use Qt to survive. And those are desktop
> developers. Show me lot of not "hello world" apps writte using qml. Where
> are they? On symbian? Maybe in MeeGo? I don't see any on the desktop. But i
> see Guitar Pro right now which is based on QWidgets.
>
> About O-notations. Article i mentioned shows that Quick-1 showed 90 FPS
> using QPainter. Why do we need to raize up performance to 250 FPS, if user
> can't see more than 24 frames per second? Excpecially in static UI. In my
> IPhone i don't see where i need such a big frame rate. I don't mention
> games (they need such FPS to freen CPU resource), and i think it really can
> be useful there. But NOT on the desktop. Almost all advantages of QML are
> gone there - batteries are big, CPU are fast. Painter works perfectly. And
> yes, i need _native_ interface here which is done using painter/styles. I
> don't want to implement 3 different interfaces for each platform. I want to
> use my old code. And you telling me to throw it away. Why the hell i should
> do that? First let microsoft deprecate winapi and reimplement whole OS from
> scratch. Then you can start telling me what to do.
> And what about KDE? It is all based on QWidgets. And it took them many
> years to make stable release for KDE 4 (KDE 4.4 was the first release i
> could use). You want them to reimplement whole KDE using QML? And spent
> next 4-5 years fixing bugs that already occured in QWidget stack? No,
> thanks. Finish QML first, create big declarative item library (with *all* functionaloty
> that present in QWidgets, fix *all *bugs and only then say that painter
> and widgets are outdated).
> Last qt releases (4.8) are crap - they give only bugs. Lot of people still
> prefer to use 4.7.4 (not me, however). As for Qt5 - even examples from Qt
> Demo crash. Even "outdated, but stable" widgets. And, which is more
> important, qt5 release is delayed.
>
> What i'm trying to say - you stop maintaining qwidgets, but giving us
> promises about "beautiful qml" for 3 (three!!) years. And it is still not
> here. I don't believe it can replace qwidgets, sorry. In qt6, maybe, but
> not now. So continue adding new widgets and fixing old bugs, please.
>
> 18.05.2012, в 10:29, <marius.storm-olsen at nokia.com> <
> marius.storm-olsen at nokia.com> написал(а):
>
> Sounds like marketing?
>
> It might be a 'constant' in the grand scheme of big-O notations. However,
> if the result is that you can only get 24 fps on low-end HW with large
> power footprint vs. 60 fps with HW acceleration, lower power footprint and
> leaving the main CPU to do more important tasks, what would you call the
> former? That's right, outdated technology. Scene graph is very much based
> on what todays graphics cards are optimized for, while QPainter.. well,
> it's not. We have gone as far as we could with QPainter, and needed
> something new to survive the next 5-10 years. It's not about marketing, I
> assure you.
>
> --
> .marius
>
>
> On 5/18/12 7:36 AM, "ext Иван Комиссаров" <abbapoh at gmail.com> wrote:
>
> Btw, you're saying that painter technology is outdated? What speedup
>
> provides QML scene graph? According to this
>
> http://labs.qt.nokia.com/2011/05/31/qml-scene-graph-in-master/ article,
>
> speedup is 2.5 times. As for me, it's just a constant optimization, it is
>
> not reduces complexity very much, as for me.
>
> If you say you reduced speed 10 times or 100 - it's the other issue. But
>
> saying that painter is outdated just because it 3 times slower than new
>
> qml... sounds like a marketing.
>
> _______________________________________________
>
> Development mailing list
>
> Development at qt-project.org
>
> http://lists.qt-project.org/mailman/listinfo/development
>
>
>
>
>
>
> _______________________________________________
> Development mailing list
> Development at qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/development/attachments/20120518/ba8979d3/attachment.html>