Thanks for pointing the Selection Service out. Linking views together is exactly what we need (and is essential to providing a rich user experience).

We’ve been tracking the development of e4, which sounds very exciting! But as you say, it’s still early days and we probably won’t be adopting it just yet (though we are using it for some prototyping etc).

All the best,

Jits

You’re right that the RCP way would be to have single instances of each view displaying the current selection (probably the editor that has focus). There is a Selection Service in RCP that facilitates this type of communication.

I don’t know that you can really achieve the UI structure you describe using RCP. RCP is great if your application fits into it’s UI model (workbench with perspectives containing editors and views). You could look at the work being done in e4 which is going to be released I think in July. e4, or the Eclipse SDK 4.0 allows for much more flexibility in the way a UI is structured. But to be honest, the new 4.0 framework might be a bit rough around the edges for mission-critical development, but depending on your needs it might be a good starting point.

— Patrick

Thanks for the response. My question reveals my naiveté when it comes to Eclipse RCP :-) You’ve mentioned some very useful info. Let me also try a different explanation…

We’re building an RCP app whereby a user can open and work with many files at once, but not in the same way users would do so right now in the Eclipse IDE (for example) – we don’t intend to show a file browser and we would like each open file to have one corresponding tab at the highest level possible in the UI (think Chrome/Chromium). This tab then has it’s own things (views?) within it that presents a complex UI to work with the file (things like editors, palettes, consoles, etc). Each tab would maintain it’s own instances of it’s internal UI bits and it’s own state, completely independent of the other tabs.

But now I think I may approaching this with the wrong concepts in mind; I didn’t fully grasp that I couldn’t add a view inside another view, so thanks for pointing that out. I guess I need to separate out the notions of “a reusable UI bit” vs. “an Eclipse View”.

It seems to me that the Eclipse RCP way of doing things would be to have common views within a perspective, that then change depending on what object is currently “selected”. Is this correct?

Thanks,
Jits

I might not sure I’m clear on what you’re trying to do. Are you asking if you can declaratively place a view inside of another view? This isn’t currently possible either programmatically or declaratively using RCP.

Or if this is about adding multiple views in the same tab group, you can do this by adding a view relative to an existing tabbed view with a “stack” relationship.

Let me know if I’m close to helping :-) Or maybe rephrase the question.

— Patrick

Whilst I agree that declaratively building layouts for perspectives makes things much more modular and extensible, it seems a bit of an incomplete solution if I can’t do the same for views within the perspective.

I hope I am missing something here, but is it possible to build something like this declaratively in Eclipse RCP:

Perspective A >> View A (tabs container) >> Instances of View B (multiple tabs which are separate instances of the same View class) >> views/layout within the Tab View B

(where the ‘>>’ indicates that I want a hierarchical/containment relationship).

It seems at the moment that I can use a perspectiveExtension to add View A to Perspective A, but then I am left to programmatically build the rest. Is this true?

Thanks in advance for any help.

Jits

You need to add a secondary id to your views when you add them in the perspectiveExtension. Instead of “myview.id”, enter “myview.id:1″ and “myview.id:2″.

Hope this helps,

— Patrick

I am struggling to add two instances of the *same* view (which has allowMultiple=true) to the same perspectiveExtension: only one view is showing.

Do you have an idea if this achieveable?

Thanks

I agree about the template and voted for the Bugzilla entry. Thanks for doing this!

— Patrick

I agree that the declarative approach is the preferable.

I adjusted therefore my Eclipse RCP tutorial (http://www.vogella.de/articles/RichClientPlatform/article.html) so that it first demonstrates the usage of perspective extension and then afterwards demonstrates the usage of code.

I believe it is tempting to stick with the things you first learn. As the Eclipse view RCP template uses code to stick the view to the perspective most of the people stay with this approach. A Bug report is open for this https://bugs.eclipse.org/bugs/show_bug.cgi?id=265231

Best regards, Lars