what you're actually getting is an instance of DateTime::Format::MySQL, except you weren't supposed to have to think about that.

Yes, I gathered that much, but shouldn't all of the classes it might return (depending on the db backend used) have a parse_datetime function, and shouldn't usage for this function be normalised across all of them and documented in DateTime::Format::DBI's documentation? It shows how to do the conversion the other direction, without caring which database is on the backend. Why not in this direction also? I was assuming that the underlying classes all simply implemented the things required by the general class in whatever manner is necessary for their respective backends. Apparently this is not the case?

---

Quidquid latine dictum sit altum viditur.