[Pgpool-hackers] Using application_name feature in 9.0

[Tatsuo Ishii]

> There is application_name for applications that take care of it and
> don't want it to be overriden, and there is fallback_application_name
> for applications who want to set up a default application name but who
> allow people to override it. For example, all the command line tools of
> the PostgreSQL distribution use the fallback_application_name startup
> parameter. pgAdmin does it too.
> 
> It would be interesting to send the fallback_application_name parameter
> as part of the connection packet if the application doesn't supply one
> by itself.
> 
> I tried to do it, but I unfortunately don't have more time to work on
> this now. I first need to work hard on my mentor's stuff for the
> pgAdmin's GSoC which will end soon. If you don't have time to work on
> this or can't work on this for some reason, I'll try to provide a patch
> for the 3.1 release.

Thank you for the explanation. It's clear now.  To implement this, we
add fallback_application_name to the startup packet if the application
does not supply it, then forward to backend. This is
straightforward. Harder part is reusing the connection. Pgpool checks
not only user name and database, but the startup packet lenght for
safefty to decide if we could reuse the connection. Apparently the
startup packet just the app sends will not same as the one pgpool
keeps inside.

Unfortunately I don't have time to implement either. I think we could
postpone it for 3.1.
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp