> >
> Yeah, I think they do that mostly to keep everyone happy. If the
> language is very vague, then it is almost impossible to create
> something that DOESN'T comply with the standard, due to the incredible
> range of possible interpretations.
> In any case, I did want to make the point that I DON'T think Postgres
> is doing this correctly right now, as I believe the examples in the
> original posting showed the NULL-valued records at the end of the
> sorted list both when sorted ascending and descending. I believe the
> standard is saying that NULLs should always be either at the beginning
> or end of sorts, and thus at OPPOSITE ends of ascending and descending
> sorts.
>
You know that interpretation thing....
If you look at other standards for interpretation, for example the JDBC
standards, there are four possibilities:
NULLs sort high
NULLs sort low
NULLs always at end
NULLs always at beginning
I think the standard says "do what you want, high or low, beginning or
end - as long as you are consistent"
Adrian
------------------------------
