Alvaro Herrera <[EMAIL PROTECTED]> writes:
> Tom Lane wrote:
>> The only thing we'd lose is that dropping a column
>> originally declared as serial wouldn't implicitly drop the sequence.
> Wasn't that the primary purpose that the main coder for dependencies did
> the work for?
My recollection is that the dependency for serials was added as an
afterthought without too much consideration of the long-term
implications. It was a cheap way of sort-of solving an immediate
problem using a mechanism that we were putting in place anyway.
But what we've got now is a misbegotten cross between the theory that
a SERIAL is a unitary object you mustn't muck with the innards of,
and the theory that SERIAL is just a macro that sets up an initial
state you can ALTER to your heart's content later.
IMHO we should make a choice between those plans and stick to it,
not add more and more infrastructure to let you ALTER things you
shouldn't be altering. Either a SERIAL is a black box or it isn't.
If it is not to be a black box, we need to reduce rather than increase
the amount of hidden semantics.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster
