Is it possible to use user defined aggregates (clr) with window functions (over)?

Question

You're right that it's tricky to find anything in the documentation. But searching the Connect website, I managed to find this gem:

Today, you can use CLR aggregates with OVER clause and PARTITION BY just like regular aggregate functions. Once we have support for window functions...

Which was a response from Microsoft.

However, searching on the Connect site was what I did whilst I was waiting for my aged machine to create a new database project and create this aggregate:

using System;
using System.Data;
using System.Data.SqlClient;
using System.Data.SqlTypes;
using System.IO;
using Microsoft.SqlServer.Server;

[Serializable]
[Microsoft.SqlServer.Server.SqlUserDefinedAggregate(Format.UserDefined,MaxByteSize = 2000)]
public struct SqlAggregate1 : IBinarySerialize
{
    private SqlString last;
    public void Init()
    {
        // Ignore
    }

    public void Accumulate(SqlString Value)
    {
        last = Value;
    }

    public void Merge (SqlAggregate1 Group)
    {
        // Ignore
    }

    public SqlString Terminate ()
    {
        // Put your code here
        return last;
    }

    public void Read(BinaryReader r)
    {
        last = new SqlString(r.ReadString());
    }

    public void Write(BinaryWriter w)
    {
        w.Write(last.ToString());
    }
}

And then run this script:

select dbo.SqlAggregate1(Column2) OVER (PARTITION BY Column1)
from (select 1,'abc' union all select 1,'def' union all
      select 2,'ghi' union all select 2,'jkl') as t(Column1,Column2)

Which produces:

------------
abc
abc
ghi
ghi

Which is a long way to say - you could have easily discovered the answer for yourself just by trying it.