Roman Rokytskyy wrote:

>Nickolay suggested interesting idea. Here it is:
>
>"There is a way to implement JDBC spec while providing positioned
>updates. Do not pre-fetch additional record during normal fetch.
>Do it in ResultSet.isLast() method instead. While implementing
>getCursorName() clearly state in its documentation that isLast()
>method moves the physical cursor (while logical position remains the
>same). Thats it."
>
>I cannot say if it is correct (seems so, but I have to try). I will
>check if we can add the code to JayBird. Then you can use it too, if
>you would care to change your brilliant implementation into something
>more ugly. :)
>
>

Perhaps I am missing something, but doesn't this mean that after a call
to ResultSet.isLast()
the current record can no longer be changed with a positional update?
If this is correct, then
Nickolay's solution isn't JDBC compliant. The purpose of any legitimate
standard is
interoperability. If a standard comforming program doesn't work with an
implementation,
the implementation isn't correct. Probably every JDBC implement cheats
or fudges
somewhere (except Sybase, whose cheats and fudges are defined into
JDBC), but for
a worthless and stupid method like isLast, what is the point? If a
programmer is so lame
as to require isLast(), what is the likelihood that he has even bothered
to read the Jaybird
documentation? Or have I missed something important?