On Thu, Oct 23, 2025 at 1:33 PM David Rowley wrote:
> On Thu, 23 Oct 2025 at 16:43, Richard Guo wrote:
> > How about using the following wording in the commit message?
> >
> > "
> > The previous logic in query_is_distinct_for() was incomplete because
> > the check was insufficiently thorough and
On Thu, 23 Oct 2025 at 16:43, Richard Guo wrote:
> How about using the following wording in the commit message?
>
> "
> The previous logic in query_is_distinct_for() was incomplete because
> the check was insufficiently thorough and could return false when it
> should have returned true.
> "
"it
On Thu, Oct 23, 2025 at 12:07 PM David Rowley wrote:
> Or if it's a case of it returning false when it could have returned
> true, then maybe the commit message should make that clear. I'm unable
> to tell from reading it. Something like; "The previous logic in
> query_is_distinct_for() was incomp
On Thu, 23 Oct 2025 at 15:59, David Rowley wrote:
> > No backpatch as this could result in plan changes.
>
> If this is broken then it'll need to be backpatched as if that
> function returns true when it should return false, then you could have
> LEFT JOINs being removed when they shouldn't or joi
On Thu, 23 Oct 2025 at 15:45, Richard Guo wrote:
>
> On Wed, Oct 22, 2025 at 6:25 PM Richard Guo wrote:
> > Attached is a patch along the lines of option #2. The LCOV report
> > indicates that there is currently no test coverage for the "else if
> > (query->groupingSets)" branch in query_is_dist
On Wed, Oct 22, 2025 at 6:25 PM Richard Guo wrote:
> Attached is a patch along the lines of option #2. The LCOV report
> indicates that there is currently no test coverage for the "else if
> (query->groupingSets)" branch in query_is_distinct_for(). This patch
> also adds test cases to cover that
While working on the fix for the push-down-HAVING optimization, I
noticed $subject.
query_is_distinct_for() is intended to determine whether a query never
returns duplicates of the specified columns. For a query using
grouping sets, if there are no grouping expressions, we know there are
no non-e
