Re: optimize hashjoin

<div dir='auto'><div><div><div class="elided-text">On Aug 21, 2024 19:17, Robert Haas &lt;robertmhaas(at)gmail(dot)com&gt; wrote:<br type="attribution"><blockquote style="margin:0 0 0 0.8ex;border-left:1px #ccc solid;padding-left:1ex"><p dir="ltr">On Wed, Aug 21, 2024 at 12:31 PM Tomas Vondra &lt;tomas(at)vondra(dot)me&gt; wrote:
<br>
&gt; Anyway, I took a look and the basic idea is simple - when shuffling
<br>
&gt; tuples between batches in a hash join, we're currently deforming the
<br>
&gt; tuple (-&gt;slot) we just read from a batch, only to immediately form it
<br>
&gt; (slot-&gt;) again and write it to the "correct" batch.
<br>

<br>
Does skipping this cause any problem if some attributes are toasted?
<br>

<br>
I suppose not, just something to think about.
<br>

</p></blockquote></div></div></div><div dir="auto">I don't see why would this cause any such problems - if anything has to be done when forming the tuples, it had to be done the first time. Shuffling tuples to a different batch may happen, but AFAIK it's really just a copy.</div><div dir="auto"><br></div><div dir="auto"><div><div dir="auto">--</div><div dir="auto">Tomas Vondra</div></div></div></div>