On Mon, Oct 27, 2003 at 06:15:55PM -0500, Dave Cramer wrote:
> It appears that this is a 7.4 issues, previous versions used to destroy
> a cursor if it existed, 7.4 does not do this.
I sent some patches to Barry last week which should fix this (among other
things they deal with fetchsize better, and regenerate the cursor name on
each query execution) -- although I don't have a 7.4 server to test against.
He was going to hold them until post-7.4 before applying, though, since they
weren't trivial changes.
> Dave
>
> On Mon, 2003-10-27 at 16:52, Dave Cramer wrote:
> > Just going through the code and the CursorFetchTest fails on the 3rd
> > iteration
> >
> > This is because the code can't deal with changing the fetch size on an
> > already declared fetch size.
> >
> > the question is do we want that ability. the spec of course is
> > ambiguous.
What's ambiguous exactly? At most the fetchsize is a hint, we should still
behave correctly regardless of what value that hint takes.
-O