mzak (7/21/2010)

---

Kevin Rathgeber (7/21/2010)

---

I have a comment regarding your T-SQL

In trying to keep all things equal, Shouldn't you declare a variable at the beginning of your function to hold the value of PI() / 180 instead of calculating it 4 times.

That was the first item I noticed.

Yep!

The point of the article wasn't to pit SQL CLR against T-SQL implementations (I debated whether to include the T-SQL results with it's current implementation in the article). Rather, it was to show that you don't have to use a sql data type in your SQL CLR (you can use native .net data types) and that doing so may result in performance gains - in my particular example, rather significant performance gains.

I figured Jeff would mention this shortly after I posted the requested T-SQL so I was anticipating it :-).

Good catch Kevin.

It should also be stated that this wouldn't make 'all things equal' per se. The SQL CLR implementation doesn't perform the calculation once, let alone 4 times with each call. By declaring a variable to eliminate the T-SQL redundancy, it still must be performed once with each call.