Frank,
I never said that. Everyone cares that's why we having this discussion
going. But different people different ideas how we should do care :)
And I do agree with Alex too, he was the one who started with very early
prototype and kick-started this thread.
I see Alex as a guy of lets-do-it rather lets-talk-about-it which is good.
You are absolutely right that comparing raw speed of execution flash vs
js is one thing, when comes to display list vs DOM Flash is the winner
again.
That's why we shouldn't make the advantage the bottleneck. Save raw
power because we will need some more later on for sure.
I do understand your points, but honestly I can't see advantage of mimic
exact same structure on the code syntax level.
Debugging? Maybe if we speak of production level of the flacon Jx
itself, but for developers it is absolutely unnecessary.
If I will forget import some class or dependency, my AS3 IDE will tell
me that before I will hit compile button.
After all, on AS3 side, if everything is done correctly the only thing I
expect from JS output is to be represented right way. And that I can
trust compiler to do the job.
Will not be diging and tweaking inside JS directly. If I would, then go
for any JS framework, why even bother translate this from different
language?
Because if during a development if something goes wrong with my web app,
in order to fix it I will back to the source of application, AS3.
Don't get me wrong. I am not moaning, you at least doing something and
showing us different aspects. I know you have big experience with that
after jangaroo development. I wish requireJS will prove to be useful but
it comes with a price. Even if we see adventage mostly for debugging
purposes for production we have to turn this into optimisation mode.
Which is another additional step and the whole system will be as strong
as the weakest link in this chain. And we have quite long already.
It would be useful to see side by side requireJS output and goog or even
vanilla and see this whole discussion in numbers.
But for this reason as Alex said let's get going. And I am not saying
don't try your way. Go for it prove your point, show us it can be done
efficiently.
More options better for us.
Happy Christmas one again!
Dan
On 12/22/2012 11:01 AM, Frank Wienberg wrote:
On Sat, Dec 22, 2012 at 12:19 AM, Daniel Wasilewski <devudes...@gmail.com>wrote:
With all respect, all the talks on this subject just convinced me more to
continue my work on BixBite project, because I can't see flex to compete in
near future.
Dan, what exactly in this discussion made you think that we don't care
about performance?
I am with Alex that there is enough time for moaning and complaining about
alleged performance problems when we have a real-world application
cross-compiled.
At the same time, I am convinced that if there is a performance bottleneck,
it will be in the rendering (canvas, DOM), not in the actual application
code. For example, our complex browser UI application only spends about 10%
CPU time in JavaScript code. All the rest is the browser computing styles,
DOM repaint / reflow, and the actual rendering.
My experiments with running cross-compiled Flash games on iOS had their
best performance boost with every iOS update, providing increased canvas
rendering speed and JavaScript performance.
-Frank-
