I am evaluating the effort of switching from GXT 2 to GXT 3. One of my areas of concern is the ability to build UI from Eclipse using WindowBuilder plugin. GXT 2 support in WindowBuilder is quite good. It doesn't look like GXT 3 is supported at the moment. Some of the components (e.g. ComboBox with its new constructor arguments) don't seem to play alone with the WindowBuilder plugin.
Can you shed some light on this issue please.
to the module fixed the ERRORS but now I am at the same level as you i guess with:
[ERROR] Rebind result 'com.sencha.gxt.theme.base.client.container.ViewportBaseAppearance' has no default (zero argument) constructors.[ERROR] Failed to create an instance of 'com.sencha.gxt.widget.core.client.container.Viewport' via deferred binding
All of the support in WindowBuilder for GXT 2 is based on code they've written within WindowBuilder itself. At least from the docs pages, I do not see a way that support can be added to WindowBuilder to add support for new custom Widgets - until that support is added, or until they make their own changes to support 3.0, widgets without default constructors will not be usable.
As in 2.x, ComboBox must have a ListStore to be of any use. A Cell or a LabelProvider is also required in 3 to describe how to render the content, rather than just a string to indicate which property to use (as 3.0 is not using the string-y reflection found in 2).
Correct - we've not had any input at all into how WindowBuilder works with GXT - to my knowledge, we've not been invited to either.
Taking a quick look at their bug tracker, it isn't clear if they intend to make it possible to build something versioned outside WindowBuilder itself to add support for other widget libraries. Closest bug i see is https://bugs.eclipse.org/bugs/show_bug.cgi?id=349692, but the lack of content makes it kinda hard to tell what is going on.
I know there is interest in getting GXT 3 support, but from what I know, that will involve getting patches submitted to WindowBuilder itself, which needs to wait at least until the 3.0 API is finalized and we've learned how to build within the app, as opposed to just building a plugin for it. If there were a plugin architecture, it'd be easier to consider keeping it up to date as the learning curve should be a little lower.
If you investigate further and find out it isn't that bad, post back, I'm sure you'll find lots of support both on and off the GXT team to getting that going.