Community
Participate
Working Groups
Created attachment 269993 [details] Use case to reproduce this issue Steps to reproduce: * Import the attached project * Open the UseCase1 diagram (make sure there is no Snap activated) * Perform an Arrange All * Check the C52 and C53 coordinates with the SiriusDebugView. * KO -> they have the same GMF coordinates Analysis: During an arrange all, the ChangeBoundsRequest move delta is constructed with absolute coordinates. The org.eclipse.sirius.diagram.ui.graphical.edit.policies.SpecificBorderItemSelectionEditPolicy.getMoveCommand(ChangeBoundsRequest) compute the new GMF coordinates by applying the delta on the current border node figure absolute bounds. But if one or several parents moved during the arrange all, the delta will be applied in the same way without considering that parent figure bounds will be different once the arrange all is over. We should compute the new border node GMF coordinates by using the parent GMF bounds and not the parents figure bounds that could have been modified.
Edit: The problem seems to come from the label of p5 which is not considered by the DBorderItemLocator
HOMOLOGATION For homologation, you will have to import org.eclipse.sirius.ui.debug plug-in and deploy it in your runtime. (for information you can get it from repo http://git.eclipse.org/gitroot/sirius/org.eclipse.sirius.git)