Utilizing data virtualization with dynamic data

[Austin]

Hey,

I have a scenario where we do not know the column names or data types until run-time.

We would like to be able to use data virtualization in this scenario.  Basically we construct the sql statement at run-time once we query the table for its metadata (columns).  We could add an orderby, since the VirtualQueriableCollectionView needs one. 

The question is: without pre-loading everything from the table into the VirtualQueriableCollectionView (which would defeat the purpose of this view) how do we wire our sql statement into a VirtualQueriableCollectionView.

We studied this example: GridView -> Databinding -> Various Data Sources example -> Dynamic Data

This example solves the issue of dynamically generating columns that aren't hard-coded in a model, but doesn't show how one could hook this type of solution into data virtualization.

[Vlad]

Hello,

 I've made for you small example project which may help you to achieve your goal in your case. 

Regards,
Vlad
the Telerik team

[Austin]

Hey Vlad,

Thanks for the sample project, it was very enlightening.  We have just run into one snag that we can't get around.  In our application we deal with a sql express / sql local db table that has a schema that is unknown until run-time.  Basically no solution that involves hard coding any kind of schema or model before compile time will work for us.

Thus, this notion of using the Entity Framework with a DbSet<tableSchemaClass> predefined will just not work for what we need. 

I've looked around on this site and ran into this:

=================================================

Run-time Schema Changes

The Schema Change API of Telerik OpenAccess ORM allows you to make runtime changes to the application's object model or database schema without recompilation of the application.

=================================================

This sounds similar to what we are interested in.  I guess what I am asking is if you can think of any other way to make this happen without using the Entity Framework since it seems to require some sort of hard-coded model.  Otherwise is there any way to make changes to an entity framework model at runtime?  (I was extremely unsuccessful at finding any information on that front)

I know delving into the entity framework is outside the realm of Telerik controls, but any info would be highly appreciated.

Edit-

After a lot of research and trying some strange things like creating a class at run-time via Reflection.Emit to use inside of DbSet I have realized that EF is simply not meant to work in this way.  The things it does with the model make this next to impossible.  I don't understand how such a simple concept as a database with a table with unknown columns can be such a huge hurdle.

 

[Vlad]

Hello,

 I'm not sure if this is possible with EF however I believe you can use LINQ to SQL instead similar to this article to achieve your goal.  

Regards,
Vlad
the Telerik team

[Austin]

Hey Vlad,

I'm not sure that article quite fits our use case.  I did some up with an idea though.  Basically it ends up looking like this:

                    items.ItemsLoading += (s, e) =>
                    {
                        List<Customer> list = new List<Customer>();
                        for (int i = e.StartIndex; i < e.StartIndex + e.ItemCount; i++)
                        {
                            Customer data = new Customer { ID = i + 1, Name = "Customer "};
                            list.Add(data);
                        }

                        items.Load(e.StartIndex, list);
                    };

Instead of using Linq here I just create my own list to pass to the items.Load function.  Since I don't have to use Linq I am not forced to have a static schema.  Thus I have the power to fill the list up in any way I want which I believe will solve my problems.  Thanks for all the input (that sample project was very helpful).