Обсуждение: A tiny improvement of psql

Kevin Wang <kevinpgcloud@gmail.com> writes:
> As you know, we can use the UP arrow key to get the previous command to
> avoid extra typing. This is a wonderful feature to save the lives of every
> DBA. However, if I type the commands like this sequence: A, B, B, B, B, B,
> B,  as you can see, B is the last command I execute.

> But if I try to get command A, I have to press the UP key 7 times.  I think
> the best way is: when you press the UP key, plsql should show the command
> that is different from the previous command, so the recall sequence should
> be B -> A, not B -> B -> ... -> A.  Then I only press the UP key 2 times to
> get command A.

This is driven by libreadline, not anything we control.  I have
seen the behavior you describe in some other programs, so I wonder
whether it's configurable.

> Another requirement is: could we use / to repeat executing the last command
> in plsql just like sqlplus in Oracle?

I'm pretty certain you can configure this for yourself with readline.

            regards, tom lane

On 12/26/23 17:36, Tom Lane wrote:
> Kevin Wang <kevinpgcloud@gmail.com> writes:
>> As you know, we can use the UP arrow key to get the previous command to
>> avoid extra typing. This is a wonderful feature to save the lives of every
>> DBA. However, if I type the commands like this sequence: A, B, B, B, B, B,
>> B,  as you can see, B is the last command I execute.
> 
>> But if I try to get command A, I have to press the UP key 7 times.  I think
>> the best way is: when you press the UP key, plsql should show the command
>> that is different from the previous command, so the recall sequence should
>> be B -> A, not B -> B -> ... -> A.  Then I only press the UP key 2 times to
>> get command A.
> 
> This is driven by libreadline, not anything we control.  I have
> seen the behavior you describe in some other programs, so I wonder
> whether it's configurable.

It is kind of something we control.  Per the psql docs, setting

     HISTCONTROL=ignoredups

will do the trick.

https://www.postgresql.org/docs/current/app-psql.html#APP-PSQL-VARIABLES-HISTCONTROL
-- 
Vik Fearing

On Tue, 26 Dec 2023 at 22:45, Vik Fearing <vik@postgresfriends.org> wrote:
> It is kind of something we control.  Per the psql docs, setting
>
>      HISTCONTROL=ignoredups
>
> will do the trick.

Yeah, the easiest "fix" (that I know of) for a user is to set
HISTCONTROL in ~/.psqlrc to ignoredups using:

\set HISTCONTROL ignoredups

But honestly, I think that should probably be made the default. I
can't really think of a reason who would actually want the current
default of "none". And while we're at it maybe there are some other
defaults in psql that are worth changing. The main ones from my psqlrc
that seem like good defaults for pretty much everyone:

\x auto
\pset linestyle unicode

And maybe fixing the major pitfall I always run into with psql: Having
ON_ERROR_STOP default to on when a script is passed in using -f/--file