Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.
Bug 306042 - [Shell] Shell not restored correctly after browser resize
Summary: [Shell] Shell not restored correctly after browser resize
Status: RESOLVED FIXED
Alias: None
Product: RAP
Classification: RT
Component: RWT (show other bugs)
Version: 1.3   Edit
Hardware: All All
: P3 normal (vote)
Target Milestone: 1.3 M7   Edit
Assignee: Project Inbox CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-03-16 12:42 EDT by Holger Staudacher CLA
Modified: 2010-03-17 15:03 EDT (History)
0 users

See Also:


Attachments
Quick fix that solves the problem (1.04 KB, text/plain)
2010-03-16 13:05 EDT, Ralf Sternberg CLA
no flags Details
new fix (5.71 KB, patch)
2010-03-17 09:59 EDT, Tim Buschtoens CLA
ruediger.herrmann: iplog+
Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Holger Staudacher CLA 2010-03-16 12:42:17 EDT
The error occurs when a maximized shell should be restored after a browser resize. Steps to reproduce:

1) Start a RAP application with a workbench (e.g. org.eclipse.rap.demo with entrypoint default).
2) Maximize the WorkbenchWindow (Shell).
3) Resize your browser (I have made it smaller).
4) Reload the app with a browser reload.
5) Resize your browser (make it bigger).

A really strange behavior shows up. The Shell resizes but the visible area is only the size of the last saved Shell size.
Comment 1 Ralf Sternberg CLA 2010-03-16 13:05:30 EDT
Created attachment 162185 [details]
Quick fix that solves the problem

The problem is that the bounds are also rendered to the client for maximized Shells. After a recent change in the order of render calls in ShellLCA, the client Shell receives a setMaximized and after this a setSpace call. This breaks the client.

The quick fix is to omit the renderBounds call for max'ed shells.
Comment 2 Ralf Sternberg CLA 2010-03-16 13:54:38 EDT
Committed a modified version of the patch. This is only a quick workaround, we still need a decent fix.
Comment 3 Tim Buschtoens CLA 2010-03-17 09:59:01 EDT
Created attachment 162282 [details]
new fix

This essentially moves Ralfs fix to the clients Shell.js where it's less intrusive. It also refactors the fix for bug 306030 to be solved in Window.js (where it belongs) instead Shell.js, and bug 306043 to actually send valid values instead of ignoring invalid ones. The qx.js must be rebuild.
Comment 4 Rüdiger Herrmann CLA 2010-03-17 15:03:52 EDT
Applied second patch to CVS HEAD