Re: PRIMARY KEY on a *group* of columns imply that each column is

Marco Colombo wrote:
> That is, given a variable A, you can always write a boolean expression
> that evaluates True or False to test if A is _equal_ to None/undef/NULL
> (in C):

And of course you can do this in SQL as well.

> ('a', NULL) is neither the same _nor different_ from ('a', 2).

Uh, I'm not sure what you mean by "the same"; but
certainly ('a',NULL) is distinct from ('a',2).

>The result of comparison is NULL, no matter how you're testing it:

Unless you're testing with the IS DISTINCT FROM operator.



fli=# select ('a',NULL) IS DISTINCT FROM ('a',2);
  ?column?
----------
  t
(1 row)

> marco=# select ('a', 2) = ('a', NULL);
>  ?column?
> ----------
>
> (1 row)
>
> marco=# select ('a', 2) <> ('a', NULL);
>  ?column?
> ----------
>
> (1 row)


fli=# select ('a',NULL) IS DISTINCT FROM ('a',NULL);
  ?column?
----------
  f
(1 row)


> That's why NULLs are not allowed in primary keys...

But in general I agree with most of the rest of what you said.