Is there a way to write a Pivot() using a LINQ query against a DataTable?

[John]

Title:

Is there a way to write a Pivot() using a LINQ query against a DataTable?

Question: I've got a bit of data problem I'm trying to figure out.


Essentially, I'm dealing with an Entity/Attribute/Value type of schema,
similar to IDS. Currently, we're sitting on an Oracle backend, but will
eventually move to something new (NOSQL, postgres,...) So, rather than push
to work to the database using a command (pivot) that might or might not be
supported, we want to do the work in code.

So, this is where we are...
When I initially query the database, I can a dataset back in the form:
NOTE:: I will


ENTITY TYPES
-----------------
ENT_TYPE_ID  ENTID  ENTNAME
======================
1                      1         a1
1                      2         a2
1                      3         a3
1                      4         a4
2                      1         b1
2                      2         b2
2                      3         b3
2                      4         b4

NOTE:: Each ENT_TYPE_ID can have 1-N ENTIDs

ID  ENTID VAL
=============
1   1     w1
1   2     x1
1   3     y1
1   4     z1       I'd like this to map to a table looking like this:
2   1     w2
2   2     x2           (Cols) ID  a1   a2   a3   a4
2   3     y2           ---------------------------          
2   4     z2                      1   w1   x1   y1   z1
3   1     w3                      2  w2   x2   y2   z2
3   2     x3                       3  w3   x3   y3   z3
3   3     y3                       4  w4   x4   y4   z4
3   4     z3
4   1     w4
4   2     x4
4   3     y4
4   4     z4

The datatable will become the datasource for an Infragistics UltraDataGrid.
Our columns & data set are pretty dynamic in nature.
We'd like to do this as officially as possible in C#.

In Flex/Actionscript (and I assume Java), there's the notion of creating a dynamic
object based on named/value pairs. This works well in decoding JSON objects.
I'm kind of trying to approximate this in C#.

The 1st thought is to go to an array of dictionaries: ISortedDictionary<key, val>.
Now the question becomes, how to get from that to a DataTable?
Even though I'm using an list of items, I don't think it'll be directly bindable.
Dictionary objects can't be directly used to bind as a data source to a grid (or list, etc).

What's the best way to massage the array/list of dictionaries into datatable?
I know I could do it the brain dead way of iterating..., but is there a slicker
better performing approach I can take?

I took a look @ Tuples in C# 4.0.  As near as I can tell , they won't work in my implementation.
I am/will be dealing with a dynamic number of entities (1->N). The system should respond & behave
accordingly when presented with any number of Entities (Ent IDs).

Is there a way to do this in LINQ? I've been able to google a bunch of examples, but they
all dealt with a fixed number of entities.  I only know the number of entities @ runtime, so I ve
need to create a LINQ query that can PIVOT() on a dynamic set of entities.

Can LINQ somehow provide a PIVOT solution like that?
 
Performance considerations could be an issue here. It's unclear to me what the volume of through put will be.

Any help would be greatly appreciated here...

[Petko_I]

Hi John,

The issue you are posing is quite interesting but I do not see how this relates to Telerik OpenAccess ORM. If you use it, you will not need to populate DataTables and use them as a data source. Besides, we have extended integration between our controls and OpenAccess which will ease the process of binding. To simulate the Pivot() with LINQ and not rely on a constant number of entities you will need to resort to dynamic LINQ. Yet, some experimentation needs to be done there so that you see what extension methods you can transform with the dynamic library.

Please, contact us right away should you need any information regarding OpenAccess.

All the best,
Petko_I
the Telerik team