Hi Solomon -

This is definitely an interesting solution. I had noticed that ugliness of having to wait for the whole recordset to render before you could send it downrange with CLR TVF's, so I will take a look at this one for sure next time it comes up.

Besides the speed - I'd anticipate this also reduces the memory footprint as well - have you had any exposure to this (it should be smaller since the entire recordset doesn't need to live in memory beofre it starts being sent)?