Thanks for the comments.
On 19/05/2011 15:46, Øystein Grøvlen wrote:
> Hi Olav,
> Thanks for the patch. I have verified that it fixes the issues. I
> have the following the questions to the patch:
> - Why is it sufficient with an assert for some of
> update_used_tables() functions? (I think the reason for that
> should be added to the commit comments.)
I am not sure if it is sufficient with an assert for these functions -
that is why I added the asserts in order to detect if it is possible
that the value of with_subselect can change for these Item types. For
the cases where I added the assert I was not able to think of cases or
hit it when running the RQG subquery grammar extensively. Anyway, since
the cost of updating this flag also for these Item types is very small I
have made an updated patch where all asserts are replaced with actual
code for propagating the with_subselect flag.
> - Did you forget Item_is_null::update_used_tables()?
Assuming it is Item_func_isnull::update_used_tables(): yes, I had
overlooked that one. Thanks for spotting it.
> I also think the commit comment for item_strfunc.cc should also be
> added to the commit comment for item_cmpfunc.cc.
The updated patch is available here: