Community
Participate
Working Groups
When an RCP application (or even the Eclipse IDE itself) is started for the first time, the window position is such that the app window extends past the screen's display area (if the app window's size is close to the size of the screen). There seems to be no way to change the default starting location of the application window. An implementation of IWorkbenchWindowConfigurer.setInitialPosition(...) would be great - it would allow an RCP application to set the default window position in the WorkbenchAdvisor.preWindowOpen(...) method, along with the default window size.
We should fix the initial window position even if we don't provide this override. I hadn't noticed it before myself, but now that you point it out I cant imagine how I missed it. Created Bug 85003 to track that defect.
A workaround is to set the shell location in postWindowCreate.
This workaround does work - but too well. This workaround makes the window _always_ appear at the coordinates given in postWindowCreate, even if the workbench state was loaded from a previous run of the app (without this workaround, the window location is restored from the previous run). What I'd like is a way to provide a default window location for when there is no workbench state to restore. In postWindowCreate I could check for the existence of a workbench.xml file to determine if I should set the shell's location, so this isn't a serious issue - but now I'm working around the workaround, so I still think a more proper way should be implemented.
Set a flag in postWindowRestore using configurer.setData, and check it in postWindowCreate.
I'm going to have to defer this new API to post-3.1, unfortunately, but we will address bug 85003.
Reassigning bugs in component areas that are changing ownership.
Prakash is now responsible for watching bugs in the [RCP] component area.
I noticed that I can set the initial location of my Trimmed Window via the Application.e4xmi, but not the size (I still have to use my ApplicationWorkbenchWindowAdvisor for that). Does that invalidate this bug? Is there another bug for not being able to set the size? I'm working on a RCP application, using the compatibility mode, with a Mars.2 target platform.
