Re: Getting a ResultSet for a refcursor element.

Barry Lind <blind@xythos.com> writes:

> But now we essentially have two types of ResultSets, regular ones and
> refcursor ones.  Refcursor ones need to be treated differently, because
> if you want to do the above the refcursor based result set can't
> populate its results until you attempt to get the results (i.e. call
> next()), this is because if you do intend to pass it on you can't get
> the results because then the called function will have nothing to fetch
> since the data will already be fetched.
>
> Also it wouldn't be appropriate to pass any old result set in as a bind
> since only refcursor result sets could be passed on.
>
> But perhaps there is a middle ground between our two view points of view
> that combines them.
>
> You are suggesting getting a refcursor returns a regular result set, I
> am suggesting it returns a special object that retains the pointer
> characteristics of a refcuror.
>
> By combining these ideas we could do the following.  Introduce a new
> object RefCursorResultSet that extends ResultSet.  It basically acts as
> my wrapper object and can be passed on to other function calls.  However
> at the same time it can also fully implement ResultSet.  So that once
> you call next() it will actually go out and fetch the data from the
> refcursor.  So I would see the implementation of this object as having a
> method like String getCursorID() that would return the refcursor name.
> It would also have a member variable that would be the real result set.
>   The implementation of next() would initialize the member the first
> time it was called and then it and all other methods from ResultSet
> would simply call the corresponding methods on the internal ResultSet
> object.
>
> How does this sound?

That sounds great. That is very like what Oracle does.

Note that I am not opposed to your idea, I think it's a good
idea. However, I have a need to keep Oracle compatibility in this
area (because I'm porting lots of code).

But if you're happy with the above, then I'll resubmit the patch on
that basis.



Nic