Re: Suggested fix for \p and \r in psql

From: Fabien COELHO <coelho(at)cri(dot)ensmp(dot)fr>
To: Daniel Verite <daniel(at)manitou-mail(dot)org>
Cc: PostgreSQL Developers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Suggested fix for \p and \r in psql
Date: 2017-04-02 14:58:38
Message-ID: alpine.DEB.2.20.1704021645010.4632@lancre
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers


Hello Daniel,

> I've noticed two issues with the query buffer post-commit e984ef5
> (Support \if ... \elif ... \else ... \endif in psql scripting):

I thought that Tom's changes were somehow intentional, in order to
simplify the code.

> 1. \p ignores the "previous buffer". Example:
>
> postgres=# select 1;
> ?column?
> ----------
> 1
> (1 row)
>
> postgres=# \p
> Query buffer is empty.

Yep. Note that it still works with:

SELECT 1 \g

> That doesn't match the pre-commit behavior, and is not
> consistent with \e or \w

Indeed, there is definitely an issue because \g executes something where
\p prints nothing...

10devel> SELECT 1; -- 1
10devel> \p -- <empty>
10devel> \g -- 1!

> 2. \r keeps the "previous buffer". I think it should clear it. Example:

I'm not that sure, \r clears the current buffer and restores the previous
one, so that two \r are needed to fully clear:

9.6> SELECT 1; -- 1
9.6> SELECT 2 \r -- first clear
9.6> \g -- 1 (redo SELECT 1)
9.6> \r -- second clear
9.6> \g -- <nothing>

Patch applies, make check ok. However:

I tend to agree that 1 above is a regression, but ISTM that 2 is somehow
the expected behavior, i.e. \r should not clear the previous buffer, just
the current one, so that two \r are needed for a "full" clear. \r should
clear the current buffer and restores the previous one.

--
Fabien.

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2017-04-02 14:59:17 Re: Performance improvement for joins where outer side is unique
Previous Message Dong Yuan 2017-04-02 14:54:50 [GSoC] Question about "Explicitly support predicate locks in index access methods besides btree"