On Saturday 21 May 2011, Thiago Macieira wrote: > On Saturday, 21 de May de 2011 12:30:17 Konrad Rosenbaum wrote: > > On Saturday 21 May 2011, Thiago Macieira wrote: > > > It should hopefully be basically a recompile for 95% of the apps. > > > > With all the announcements of changing the infrastructure of Qt, like > > changing from widgets to QML, from raster based painting to OpenGL, I > > somehow have doubts about both statements. > > There's a difference between focus and what will be there. The current > widget classes will be there, so the current applications that use them > will still compile and work. > > However, we want QML to also work for the desktop and that's where the > energy will be spent in.
My point was: I'm not certain there is enough energy to make that work and
make it work well within one year.
[cut]
> The above two statements imply that we want pure C++ UIs. That's not the
> design goal. You may disagree and the goal may change, but right now,
> it's not the goal.
Let's check on the summit whether there are more fans of QML or more fans of
old style C++ widgets...
I'll shut up and buy you a beer if you are right about us die-hard widget-
lovers being an insignificant minority.
> > * QML does not have a native look on Desktops yet, the classes that are
> > supposed to do that do not seem to be in alpha, let alone beta, state
> > yet
>
> Indeed. They will most likely not be ready before Qt 5.0. So?
So I'm concerned. It would be a show-stopper for 5.0. (No surprise in either
statement.)
> > * there seems to be at least some confusion about whether QML and
> > widgets can be freely mixed
>
> Only research will tell. If they can't, what's the problem?
Then the porting of existing programs to new display paradigms becomes
harder by at least an order of magnitude because you cannot mix existing
special purpose widgets into new windows or new special QML elements into
existing dialogs. Instead of a gradual transition we would be forced to
rewrite those windows/dialogs.
> > * it is completely unclear how stable rendering will be provided on
> > hardware with unstable OpenGL drivers (most non-professional PC
> > hardware)
> >
> > * there is no viable driver for platforms without OpenGL (many embedded
> > systems, framebuffer based rendering, Windows with minimal DirectX,
> > etc.); Trolls are confident that a wonder will happen all by itself
>
> Indeed. Those are major points to be looked into, otherwise Qt 5.0 will
> not run on that hardware, which is contrary to its stated goals.
So, is there a contingency plan?
> > * it is unclear how difficult it will be to port widgets that do their
> > own custom rendering (using QPainter)
>
> Rewrite using QML.
Ouch. I do not think my customers will be happy about the prospect of me
spending weeks on things that were already done in previous releases.
> QPainter will be there and will render onto an off-screen surface, which
> the GL engine will take and display at the right place, though.
Good, so I can continue to use them as widgets. If they could be mixed with
QML I could use them in newer code.
> > * some parts of the library that people rely on are in deprecated state
> > (e.g. QtXml), we will see in June whether they get picked up by the
> > community
>
> And if they don't, they will remain as they are exactly now.
Ok.
Konrad
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Opengov mailing list Opengov@qt-labs.org http://lists.qt-labs.org/listinfo/opengov
