Pretty good read! But I don't like how the author tries to characterize generic ORM's like Django's as "Portability at all costs". I don't think the point is that you can switch DB's a few years down the road if you choose. He's right, nobody does that. Ever :) The advantage is that you can start a new project with whatever DB you already have, or are comfortable with. So, if you want to begin a new Django project, you aren't locked into using MySql, or Postgres, or what-have-you. It is true you sacrifice being able to leverage the full gamut of features a particular engine might offer, but that is a compromise you make with any ORM right? Or, if the limitations are to much to bear, just don't use Django ORM, it's not required, it's just the path of least resistance.
