There have been reports that the performance of the DMN editor's grids does not scale. On examination performance for anything other than simple examples is pretty dire.. something is seriously wrong.

On closer inspection the following issues were found:

• Multiple NodeMouseDownHandler, NodeMouseMoveHandler and NodeMouseUpHandler are registered because multiple DMNGridLayer are (incorrectly) instantiated due to a flaw in GridLienzoPanel and GridLienzoScrollHandler.
• Calls to GridWidget.getHeight() and GridRow.getHeight() can be very expensive for due to the dynamic calculation in ExpressionEditorGridRow and LiteralExpressionGridRow rows.

Manual Acceptance test

Environment: wildfly14 with following memory settings: JAVA_OPTS="-Xms64m -Xmx1024m -XX:MetaspaceSize=96M -XX:MaxMetaspaceSize=1024m -Djava.net.preferIPv4Stack=true"

GDT

• Merge view
• Pinned mode
• Change column width, save and reopen
• Search the table
• Scroll the table
• Edit the cell
• Reorder columns
• Reorder rows

DMN decision table

• Save and reopen not able to save in reasonable time
• Change column width
• Search the grid
• Scroll the grid
• Reorder columns
• Reorder rows
• Adding column
• Removing column
• Edit the cell not able to invoke edit mode by double click, if the table was scrolled edit-cell.webm