>>And if I read your article correctly, the other annoying thing is that the referential integrity builder will always disallow to set a field to null in the child table, even if the field is defined as nullable ; only the external TaxRI builder would allow this. Right ?

>I wrote that? I never tried to set fields to null. But TaxRI does have some additional options. And, the generated code is about 10 times more compact (and readable).

That is what I understood from this sentence in your article :
>This program offers some interesting additional options. One option that I find very interesting is "Allow blanks", for the "Insert" case. The idea is that you can leave a field blank, but if you write a value, the referential integrity functions do the verification in the other table, to see whether it is a valid value.

The possibility to have referential integrity on nullable fields is important to me, so this alone would already be a reason to research for TaxRI (I have not found it yet).