Peter Eisentraut wrote:
> Adriaan van Os wrote:
>>>> However, min(VARCHAROID) and max(VARCHAROID) return TEXTOID as a
>>>> result type.
>>> Yea, they are internally treated as very similar types.
>> But "internally treated as very similar" is still not "same as
>> argument type". Computing requires exactness.
>
> Aside from the apparent discrepancy between the documentation and the
> actual behavior, is there an actual use case where this is a problem?
Dijkstra's "Rule 0: Dont Make a Mess of It <http://www.cs.utexas.edu/users/EWD/> and the virtues
of strong typing, which, for SQL, imply checks at runtime <http://en.wikipedia.org/wiki/Type_safety>.
Besides, the question is absurd. I stumble over a stone on the road, report it and then you ask "is
there an actual use case where this is a problem". Why else do I report it ? What you probably
wanted to ask is: "Apart from the missing warning along the road, couldn't you have walked around
that stone ?"
Well, in answer to that last question, I could have, but that is the wrong approach to computing.
Regards,
Adriaan van Os
