Understanding hierarchical grid structure

There are two general steps to implement a hierarchical grid:

1. Design the hierarchy In this step, using the designer or code-behind you should specify the structure of the hierarchical tables. This is related to populating the collection of DetailTables of RadGrid MasterTableView and the corresponding columns. It is important that this step is completed before Telerik RadGrid is bound to any data. For example the grid can have a similar structure:

   • MasterTableView
     • DetailTable(0)
       • DetailTable(0, 0)

   Each detail table (including the MasterTableView) is control of type GridTableView. This means that each control is represented by HTML table (runtime) object. This table has RenderColumns, Columns andAutoGeneratedColumns collections (see Getting familiar with the server side API ) properties for settings style, group and sort expressions, etc.

   The whole structure of Telerik RadGrid and its detail tables is saved in the ViewState. Therefore if you build a grid programmatically, you can safely do this on PageInit or Page_Load (see Programmatic grid creation ). For illustration on defining the groups in code-behind check out this article.

2. During runtime (when data-binding) Telerik RadGrid builds special structure of items corresponding to the structure of the detail tables. For each item in the master table a new control (GridTableView) will be created and will be assigned as a child of the master item. This control will be a copy of the corresponding table - that should be DetailTable(0). Hence, if you have 3 items in the MasterTableView in the second level of hierarchy the grid will look like this:

   • MasterTableView
   • Item(0)
     • DetailTable0(copy 0)
   • Item(1)
     • DetailTable0 (copy 1)
   • Item(2)
     • DetailTable0 (copy 2)

   Here Item(0) is a parent item for DetailTable(0). You can obtain its instance using the GridTableView.ParentItem property. As there is a third level of hierarchy (note each of the items of the copies of DetailTable(0) will have a detail table) which is a copy of DetailTable(0, 0).

   For example, let's say that the copy of DetailTable(0) with index 0 has 2 items:

   • DetailTable0 (copy 0)
   • Item(0)
     • DetailTable(0, 0) (copy 0)
   • Item(1)
     • DetailTable(0, 0) (copy 1)

   Each copy of the detail tables should be data-bound to populate its items. Telerik RadGrid delegates full control of how this can be done to the developer - you should filter or select just the records that correspond to each detail table. Telerik RadGrid only restricts when this will happen - either through Declarative hierarchy relations or DetailTableDataBind event.

   If you choose the second approach, Telerik RadGrid fires DetailTableDataBind for each DetailTable that is about to be data-bound. However, this depends on the HierarchyLoadMode property of the corresponding GridTableView:

   • HierarchyLoadMode is ServerBind - the DetailTableDataBind event will be fired immediately after the corresponding parent item is bound.
   • HierarchyLoadMode is ServerOnDemand - the detail table will be bound as soon as its parent item is expanded.

   Inside the DetailTableDataBind event handler you can determine which detail table is about to be data-bound by using its DataMember property. This property should have unique value assigned in step 1.

See Also

• Overview
• Hierarchy load modes