>Not disagreeing with the theory btw - big fan of Codd myself, but reality rarely conforms to ideals. On the contrary, you should disagree with Codd's “theory” (about the specific topic of nulls—not about the Relational Model, of course!), which is untenable, hence far from ideal (on the other hand, the problem he was trying to address is inherently intractable, as it entails detecting tautologies). The reality is that SQL conforms really really well to what Codd proposed, despite his proposal being incoherent and, to his own admission, “preliminary and in need of further research”. For the curious ones: read [Codd's paper](https://web.archive.org/web/20030921174044/http://www.cc.gatech.edu:80/classes/AY2003/cs6411_spring/p397.pdf), §2.3, and compare the definitions therein with the behaviour of y̵o̵u̵r̵ ̵f̵a̵v̵o̵r̵i̵t̵e̵ ̵s̵y̵s̵t̵e̵m̵ SQLite :)