Re: pgbench - add \if support - Mailing list pgsql-hackers

From Pavel Stehule
Subject Re: pgbench - add \if support
Date
Msg-id CAFj8pRD0ZOWRKBBetFxOgLXZx5_L8jU2YmWdiuNwqSn-dKur8A@mail.gmail.com
Whole thread Raw
In response to Re: pgbench - add \if support  (Fabien COELHO <coelho@cri.ensmp.fr>)
List pgsql-hackers


2018-01-22 10:45 GMT+01:00 Fabien COELHO <coelho@cri.ensmp.fr>:

few scripting features doesn't mean scripting language. \if in psql is nice
feature that reduce duplicate code, unreadable code, and helps with
deployment and test scripts. pgbench and psql should to have similar
environment - and I am thinking so \if should be there.

Using Lua is not bad idea - in psql too - I though about it much, but in
this case is better to start from zero.

Yep. Having another versatile (interactive) client would not be a bad thing. I'm still wondering how to conciliate any scripting language with "bare SQL". The backslash-oriented syntax already used for psql & pgbench seems the only available option. Otherwise ISTM that it is back to a standard library oriented client access with import, connect, exec... whatever set of function already provided by standard libraries (psycopg for python, ...).

The implementation of some parts in C is frustrating - mainly tab complete. There is not possibility to create own backslash command - or enhance buildin commands. Is not possible to customize output.

So some hypothetical client can be implemented like some core C module - for fast processing of tabular data and all other can be implemented in Lua. I can imagine so this client can support some input forms, for bar menu, for some simple reports. It can be more like FoxPro client than command line only client. In few years we can use ncurses everywhere, and then there are possibility to write rich TUI client.

Regards

Pavel


--
Fabien.

pgsql-hackers by date:

Previous
From: Daniel Gustafsson
Date:
Subject: Re: [HACKERS] Support for Secure Transport SSL library on macOS as OpenSSL alternative
Next
From: Dean Rasheed
Date:
Subject: Re: stricter MCV tests for uniform distributions (was Re: MCV listsfor highly skewed distributions)