What is the performance expectations for Ext GWT 3.0? Better or the same as 2.0? The architecture is vastly different (no lazy rendering, heavy use of generators, cell widgets, etc.), so I'm not sure what to expect.
We are just starting the transition now, but we are focusing on new development rather than porting existing code. I'm glad that there is some end-user benefits for the switch. Feature-wise GXT 3 doesn't offer much new, and requires large code changes and retraining for developers.
That's right, but it's really worth the effort.
I'm quite enthusiatic especially because of the good performance (as this has been a great issue in all our projects).
Developing new can be a good idea, as the code changes are really very much.
I think the most important thing is our change from relative positioning of the widgets to absolute.
Before the change to GXT 3 we had to put all widgets in panels that have been placed in panels that have been placed in panels and so on, to get the correct structure of our masks.
Now we can do it without those panels. This and the fact that the labels of the widgets (e.g. text box) are part of the widget means that we have much less widgets to send to the browser AND to render (about one third)!
Additionally (I think) there is an architecturally improvement in creating the appearance of the widgets, as the style information in css are linked to the widgets at compile time (but I think one of the Sencha people can say this more detailed).