Infralution Support

[ShailendraBaghel]

I have a class which represents a Row and i have the following property to get Child objects

public Array Childs
{
get
{
return Array of child Row objects.
}
}

Row.ChildProperty="Childs"
Row.ChildPolicy=Infralution.Controls.VirtualTree.RowChildPolicy.LoadOnExpand;

In our application we can have a million child objects of a parent.
Now creating million child objects when user expands the row is not a good design as it is time consuming.
It might happen that user just sees the first 10 child objects and then scrolls down to view the last 10 child objects.

Is it possible that when user expands a row, memory of only first 100 child objects is allocated and when user scrolls down to view the next 100, then those child objects are allocated and so on?

Some psuedo-code will be helpful.

Thanks in advance.

[Infralution]

The problem is that you are using an array as the collection to hold your child objects. This means that you will have to load the entire array upfront - requiring a large amount of memory and time to load.

What you need to do is to write your own collection class that implements the standard IList interface in a more intelligent way. The two main methods of this interface that you need to implement are:

1. Count – should return the total number of items in the collection.

2. Item(index) – should return the item at the given index.

Typically you might use a hashtable to act as cache to hold items you’ve already loaded (keyed by the collection index). So your Item method would first check the cache to see if you’ve loaded the item. If you haven’t then you would load the item at the given index from file/disk, store it in the cache and return it.

If you are getting data over a network (or other source where there is latency) you can improve performance by implementing a paging mechanism so that surrounding items are also loaded into the cache. This will reduce the number of round trips to get data.

If you implement a collection like this that handles loading data on demand then Virtual Tree will use the interface intelligently and only call the Item method (triggering load on demand) for those items it actually needs to display. This means for the example you gave that Virtual Tree will call the Count method to check how many children the parent has and then call the Item method for each item currently displayed.

The advantage of implementing the on demand loading in the collection rather than have Virtual Tree attempt some type of callback scheme is that the IList interface provides a ready to use interface that can nicely encapsulate this type of functionality and other parts of your application can then make use of the on demand loading mechanism.
_________________
Infralution Support