Re: Removing useless DISTINCT clauses

From: David Rowley <david(dot)rowley(at)2ndquadrant(dot)com>
To: Melanie Plageman <melanieplageman(at)gmail(dot)com>
Cc: PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>, David Rowley <dgrowley(at)gmail(dot)com>
Subject: Re: Removing useless DISTINCT clauses
Date: 2018-03-22 05:58:32
Message-ID: CAKJS1f_KPr355Mxh+iy7GW586MN4=BTP2HUobc5Z83nct4=mTQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 21 March 2018 at 16:29, Melanie Plageman <melanieplageman(at)gmail(dot)com> wrote:
> For a small performance hit but an improvement in readability, the length check
> can be moved from the individual group by and distinct clause checks into the
> helper function
>
> if (list_length(parse->distinctClause) < 2)
> return;
>
> and
>
> if (list_length(parse->groupClause) < 2)
> return;
>
> can be moved into `remove_functionally_dependent_clauses`

I remember thinking about this when writing, and I think I ended up
doing the check earlier as I thought that having 1 GROUP BY clause
item would be the most common case, so it seems like a good idea to
abort as early as possible in that case. That's most likely not the
case for DISTINCT.

I imagine it probably does not make that much difference anyway, so
I've moved it into the remove_functionally_dependent_clauses()
function, as mentioned.

> The main helper function that is added `remove_functionally_dependent_clauses`
> uses one style of comment--with the name of the function and then the rest of
> the description indented which is found some places in the code,
> /*
> * remove_functionally_dependent_clauses
> * Processes clauselist and removes any items which are deemed to be
> * functionally dependent on other clauselist items.
> *
> * If any item from the list can be removed, then a new list is built which
> * does not contain the removed items. If no item can be removed then the
> * original list is returned.
> */
>
> while other helper functions in the same file use a different style--all lines
> flush to the side with a space. I'm not sure which is preferred

You might have a point, but remove_useless_groupby_columns already
does it this way, so I don't think changing that is a good idea.

> The new status of this patch is: Waiting on Author

Thanks for reviewing this. I've attached an updated patch. I'll set
back to waiting for review again.

--
David Rowley http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

Attachment Content-Type Size
remove_useless_distinct_clauses_v4.patch application/octet-stream 12.3 KB

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Pavan Deolasee 2018-03-22 06:29:14 Re: Faster inserts with mostly-monotonically increasing values
Previous Message Ashutosh Bapat 2018-03-22 04:36:07 Re: [HACKERS] Partition-wise aggregation/grouping