MIME-Version: 1.0
References: <17709-4f557160e3e8ee9a@postgresql.org>
 <CAMbWs48=JwUZdGRBdfSPVgovyqq=QTH8qQXoaDsfQymT3xwBNQ@mail.gmail.com>
 <CAApHDvqCT4E0A1pt86sbGwWLRhdiWcuUHx8oAq1T+LqpOMskQg@mail.gmail.com>
In-Reply-To: 
 <CAApHDvqCT4E0A1pt86sbGwWLRhdiWcuUHx8oAq1T+LqpOMskQg@mail.gmail.com>
From: Richard Guo <guofenglinux@gmail.com>
Date: Fri, 9 Dec 2022 22:01:21 +0800
Message-ID: 
 <CAMbWs4-4L6Emr4e-npde9h5M56wUFt6GwrPykAM9GD+W9cDfTA@mail.gmail.com>
Subject: Re: BUG #17709: Regression in PG15 with window functions -
 "WindowFunc not found in subplan target lists"
To: David Rowley <dgrowleyml@gmail.com>
Cc: makhmutov@gmail.com, pgsql-bugs@lists.postgresql.org
Content-Type: multipart/alternative; boundary="0000000000004ceb5e05ef659757"
Archived-At: 
 <https://www.postgresql.org/message-id/CAMbWs4-4L6Emr4e-npde9h5M56wUFt6GwrPykAM9GD%2BW9cDfTA%40mail.gmail.com>
Precedence: bulk

--0000000000004ceb5e05ef659757
Content-Type: text/plain; charset="UTF-8"

On Fri, Dec 9, 2022 at 7:53 PM David Rowley <dgrowleyml@gmail.com> wrote:

> The additional thing that seems to cause the reported error is that
> once the subquery is pulled up, the run condition also needs a round
> of constant folding done. See subquery_planner() around line 827.  The
> problem is that the target list's WindowFunc ends up with count(1)
> over .., but the run condition's one is left as count(case 1 when 1
> then 1 else null end), which preprocess_expression() will fold into
> the same as what's in the target list.


Yes exactly. That's what we also have to do.  I was debugging with a
simplified version of the query with the WindowFunc as count(t1.a) over
(...) and did not realize constant folding is also needed for the
runCondition.


> I'm now wondering if WindowClause.runCondition should be of type Node
> * instead of List *. I'd have imagined I should be passing the type of
> EXPRKIND_QUAL to preprocess_expression's type, but canonicalize_qual()
> does not like Lists.


I'm not sure about this.  From how the runCondition is constructed in
find_window_run_conditions, it seems there is no need to canonicalize
it.

Thanks
Richard

--0000000000004ceb5e05ef659757
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr"><br></div><div class=3D"gmail_quote"><div=
 dir=3D"ltr" class=3D"gmail_attr">On Fri, Dec 9, 2022 at 7:53 PM David Rowl=
ey &lt;<a href=3D"mailto:dgrowleyml@gmail.com">dgrowleyml@gmail.com</a>&gt;=
 wrote:</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px =
0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
The additional thing that seems to cause the reported error is that<br>
once the subquery is pulled up, the run condition also needs a round<br>
of constant folding done. See subquery_planner() around line 827.=C2=A0 The=
<br>
problem is that the target list&#39;s WindowFunc ends up with count(1)<br>
over .., but the run condition&#39;s one is left as count(case 1 when 1<br>
then 1 else null end), which preprocess_expression() will fold into<br>
the same as what&#39;s in the target list.</blockquote><div>=C2=A0</div><di=
v>Yes exactly. That&#39;s what we also have to do.=C2=A0 I was debugging wi=
th a<br>simplified version of the query with the WindowFunc as count(t1.a) =
over<br>(...) and did not realize constant folding is also needed for the<b=
r>runCondition.<br></div><div>=C2=A0</div><blockquote class=3D"gmail_quote"=
 style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);p=
adding-left:1ex">
I&#39;m now wondering if WindowClause.runCondition should be of type Node<b=
r>
* instead of List *. I&#39;d have imagined I should be passing the type of<=
br>
EXPRKIND_QUAL to preprocess_expression&#39;s type, but canonicalize_qual()<=
br>
does not like Lists.</blockquote><div>=C2=A0</div><div>I&#39;m not sure abo=
ut this.=C2=A0 From how the runCondition is constructed in<br>find_window_r=
un_conditions, it seems there is no need to canonicalize<br>it.<br><br>Than=
ks<br>Richard<br></div></div></div>

--0000000000004ceb5e05ef659757--