[jdev] XMPP Ping/Keepalive: Recommended method ?

[ennova2005-jabber at yahoo.com]

Assuming (for legacy reasons), the only injection point available in the code for this uni-directional "keepalive" is a message or presence packet, how "bad" is it to send a <presence /> to the server  (vs. say a jabber:iq:version or jabber:iq:time) ?

Will most servers ignore this safely and will it be considered a no-op ?



----- Original Message ----
From: ennova2005-jabber at yahoo.com
To: jdev at jabber.org
Sent: Sunday, June 18, 2006 10:47:19 PM
Subject: [jdev] XMPP Ping/Keepalive: Recommended method ?

Given that the protocol itself does not seem to have a defined keep-alive element, what is the recommended way for a client to keep its connection alive to a XMPP server ?

Can someone provide an exact wire representation of the "space keep alive" method that will not break current xmpp servers  ? ( I tried to observe the debug output of a couple of popular xmpp clients without much luck)

This is for an ad-hoc client that open a raw tcp socket and sends a few things ( it is not using a full xmpp client library)   There is only one instance of this client so we were otherwise thinking of using  jabber:iq:time or jabber:iq:version 

Are there other low overhead no-op packets we could be sending instead ? At the moment I am interested only in sending some valid traffic for this purpose (and not quite looking for  a valid response as long - as it does not terminate the connection)




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.jabber.org/jdev/attachments/20060622/54cfa891/attachment-0002.htm>