Thanks for the info. So the implementation is functional, but does so
by preventing cursors. Unfortunately, I am dealing with a large result
set, so I need to use cursors. I started modifying the jdbc driver.
Any other suggestions?
thanks,
Indra
Oliver Jowett wrote:
> Indra Heckenbach wrote:
>
>> Hi,
>>
>> After reviewing the JDBC API, I thought absolute() was intended to
>> reposition the cursor across the entire result set. Looking at the
>> pgjdbc implementation though, it appears to be positioning the cursor
>> in the cached results, which may be a subset of all results if
>> setFetchSize() != 0. I assume the implementation should be doing
>> "FETCH ABSOLUTE pos". If this is true, comparable methods are also
>> not correctly implemented for this case. Is this an issue, or am I
>> misinterpreting the API?
>
>
> You need to be using a scrollable resultset for absolute() to be
> available (although it looks like the current driver does not enforce
> this). If a scrollable resultset is requested, the driver disables the
> use of cursors regardless of the current fetchsize, so the "cached"
> resultset is always the complete resultset and there is no problem.
>
> This behaviour is an interim measure until a proper implementation
> that deals with scrollable resultsets and cursors correctly gets done.
> There are some backend issues to sort out first -- some (non-obvious
> set of) queries produce forward-access-only cursors, IIRC.
>
> -O
>
