If you want to compare apples with apples, compare the *newest* version of Postgres (8.3) with the *newest* version of MySQL
Cool idea. Why not get to work on it? Myself, I'd love to spend the next year working on database benchmarks and reliability tests just so I can have an informed opinion in a geek debate. Of course, no one will actually pay any attention to you, and by the time you're done the databases you're studying will have changed some more, but what the hell.
and ignore the non-inno backends.
You mean, pretend that most people aren't using the default?
It does surprisingly well.
Ah, so you do have some data? You've compared the results with both databases, then? What was surprising about your results?
Pg still has so much room for improvement, and it certainly isn't the easiest of the two to develop in.
And where would that "room for improvement" be, precisely? I can't imagine why you'd find it harder to develop for than any other database.
(Department of cheap irony: perlmonks.org hit a "database error" when I first tried to post this.)