Re: [PATCH] psql: add size-based sorting options (O/o) for tables and indexes

On Thu, 27 Nov 2025 at 08:52, Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> wrote:
>
>
>
> st 26. 11. 2025 v 14:01 odesílatel Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> napsal:
>>
>> Hi
>>
>> st 26. 11. 2025 v 13:44 odesílatel Euler Taveira <euler(at)eulerto(dot)com> napsal:
>>>
>>> On Wed, Nov 26, 2025, at 4:48 AM, M.Atıf Ceylan wrote:
>>> > Hello,
>>> > This patch adds two new meta-command modifiers for \dt(+) and \di(+):
>>> >
>>> > - O : sort by total relation size descending
>>> > - o : sort by total relation size ascending
>>> >
>>>
>>> Thanks for your contribution. Register your patch in the next commitfest [1] so
>>> we don't loose track of it.
>>>
>>> I didn't look at your patch but I was wondering if a general solution isn't a
>>> better way to add this feature. I wouldn't modify these specific psql
>>> meta-commands, instead, I would add a new psql meta-command that defines this
>>> property for all objects if applicable.
>>>
>>> \sort [ name | size [ asc | desc ] ]
>>>
>>> I thought about a list to be cover other sort cases too but if things starting
>>> to be complex, it is time to write your own query.
>>
>>
>> It is big question - if there should be specialized metacommand, or just variable or \pset setting
>>
>> it can be
>>
>> \set PREFERRED_ORDER size_desc
>> \pset preffered_order size_desc
>>
>>
>>>
>>>
>>> With a parameter, it appends the ORDER BY clause in the SQL commands executed by
>>> psql if applicable. Without a parameter, it uses the current behavior.
>>
>>
>> There were a lot of proposals related to this topic some years ago. I wrote a lot of variants of this patch
>> Generic design is very big, and solutions like proposed are not generic :-). We talked about this feature for maybe more than one year, and we didn't find a generally acceptable design.
>
>
> https://www.postgresql.org/message-id/CAFj8pRAVV2TFHsFCV=c9Aaeq7kPWGQBLkOwGronpAN583zqhWg@mail.gmail.com
>
>
>>
>>
>> At the end I wrote pspg, and the sort can be done (over result) there. Using a vertical cursor (column cursor) is very natural and user friendly.
>>
>> https://github.com/okbob/pspg
>>
>> Regards
>>
>> Pavel
>>
>>>
>>>
>>>
>>> [1] https://commitfest.postgresql.org/57/
>>>
>>>
>>> --
>>> Euler Taveira
>>> EDB https://www.enterprisedb.com/
>>>
>>>

Hi hackers.

I noted that this patch cf entry has the status "Ready for committer"
[0]. I do not think so. I see major design concerns in the proposal.
For my 2c, I would vote for general-purpose separate \sort command or
some suffix for meta-command as proposed by Pavel in thead from 2017.

I also suggest to rename commitfest entry to describe "what" instead of "how"

[0] https://commitfest.postgresql.org/patch/6258/

--
Best regards,
Kirill Reshke