[standards-jig] JNG Ramblings.

Jonathan Pobst jpobst at utk.edu
Fri Aug 9 18:03:16 UTC 2002


You'll have to excuse my likely uneducated question, but why wouldn't
something like this work just as well?

<route to='someone at jabber.org' from='me at jabber.org' xmlns='jabber2:message'
size='123'/>

<route to='someone at jabber.org' from='me at jabber.org' xmlns='jabber2:message'>
  <message type='chat'><body>Hey There!</body></message>
</route>

The first message is basically the envelope stuff that is being proposed,
which is then followed up the actual message.

This seems to accomplish the benefit of not having the server parse the
actual message.  It would read the first message (which it would have to do
for the envelope stuff anyways in the 'binary wire' stuff).  It would then
know the size to read off the socket and where to send it without any more
XML parsing.  Binary data could also be done the same way using a namespace
of say "jabber2:binary".

To allow compatibility with people who want to use telnet, the first message
would be optional (though the vast majority of clients would support it
through libraries or whatever, so we would get the speed boost 99.9% of the
time)  Yes, the server would have to fully parse those telnet messages, but
its going to have to to allow telnet under any scenario.

This seems (to uneducated me), that it gets the benefit of the packet
framing while still allowing a completely xml environment.  Yes, the first
route can be looked at as redundant, but its still less or equal to the
envelope stuff.


pobst




More information about the Standards mailing list