Actually, Wolfram does give me an exact closed form solution of the double indefinite integral with the assumption that all values are real. However, it's just way to complicated for me to bother using it. Here it is.

We could make a light shader that implements this equation. However, it would require quite a few nodes. Poser doesn't have an arctan function, so that one term would have to be expanded as a Taylor series. And - it would still only be correct for points in space that lie along the central axis of the light source. Other points would be wrong. It also assumes the lit surface is pointing right at the light source. If the normal is pointing elsewhere (as it does 99.99% of the time) the equation becomes even more ridiculous.

---

Renderosity forum reply notifications are wonky. If I read a follow-up in a thread, but I don't myself reply, then notifications no longer happen AT ALL on that thread. So if I seem to be ignoring a question, that's why. (Updated September 23, 2019)