[ejabberd] Preparing for version 1.1.2

Mickael Remond mickael.remond at process-one.net
Sun Sep 24 13:46:12 MSD 2006


* Badlop <badlop at gmail.com> [2006-09-24 11:33:40 +0200]:

> 2006/9/24, Mickael Remond <mickael.remond at process-one.net>:
> >This bug exists in 1.1.1 but should normally have been fixed in 1.1.2:
> 
> I've just tried again with 1.1.2, and it seems to be fixed :)
> 
> Since I tested several versions (1.0.0, 1.1.1, 1.1.2, svn), maybe I
> forgot to reduce the timeout on 1.1.2.
> 
> BTW, the timeout is default to 5 minutes. JWChat has configurable poll
> intervals of 1 second up to 30. Is there any point in having a timeout
> of 300 seconds? Shouldn't 35 seconds, or 60, be more appropiate?

I thought the same, but I have discussed this with Alexey and and told
that a very big timeout is necessary to deal with bad link quality.
That's why we have let it at 5 minutes.
However, I think it make sense to make this an option to override the
timeout value from the config file directly.

> Curiosity: This is the log when closing a JWChat client on
> ejabberd+Yaws 1.64. Even with that Yaws ERROR, the session is closed
> correctly after the timeout:

Yaws says there is an error, because he tries to push something back to
the client socket, whereas there is no more remote peer.
However, this is not a real "error" as from Yaws sysadmin point of view,
the system is working correctly. This is a "client-side error" and
should probably be reported as an info.
The same happens when you download a big file and interrupt the download
from the client side. 

I will submit the idea to change this to the Yaws mailing list.

-- 
Mickaël Rémond
 http://www.process-one.net/


More information about the ejabberd mailing list