Re: Result Set Cursor Patch

Kris,

Here it is again, I think my IDE reformatted a few classes which caused
this patch to be bigger then it needed to be. Here it is in its current
state.

I took the suggestions from the list and implemented the movement of
position using the MOVE command. This made finding the result set size
MUCH easier. In addition, I moved all the code related to chunks into a
new class called ResultChunk which lives in core.

Attached is a tarball which contains the new file as well as the patch.
This patch uses diff -c rather then diff -u as it was brought to my
attention as being the preferred format for this project. Let me know
if this is not correct.

If the formatting is major problem, I can attempt to reformat it to
desired settings. I will need to know what the "preferred" way is. I've
noticed there are several different formats used within these classes,
I could easily run the tree though Jacobe and clean it up. However,
that may be too big for the list. :)

There is partial support for reverse fetching in this patch, but it
does not work yet. In this version it is completely disabled so even if
reverse is requested it will ignore the request and use forward
fetching instead.

I will fix this when I have some extra time. I don't think the current
code supports reverse fetching either, so I don't think any current
functionality is lost.

The code attached does pass the entire test suite without problem.

Please let me know if there are any problems.

-Andy



On May 4, 2004, at 1:35 PM, Kris Jurka wrote:

>
>
> On Tue, 4 May 2004, Andy Zeneski wrote:
>
>> I submitted an updated patch to this list on yesterday, but I never
>> saw
>> my post come through. Did I just miss it or was it lost in transit
>> somewhere?
>>
>
> I haven't seen it and it's not in the archives.  If it was large it
> could
> have gotten killed.  The list has a size limit somewhere in the 30-50k
> range.
>
> Kris Jurka
>
> ---------------------------(end of
> broadcast)---------------------------
> TIP 9: the planner will ignore your desire to choose an index scan if
> your
>       joining column's datatypes do not match