Re: AOP: openh323gatekeeper
On Tue, May 29, 2001 at 07:38:34PM +1000, Hamish Moffatt wrote:
> On Tue, May 29, 2001 at 01:29:18AM +0200, Bernd Eckenfels wrote:
> > On Mon, May 28, 2001 at 08:59:27PM +1000, Martijn van Oosterhout wrote:
> > > Like the previous one, it works for me, although I havn't tested it much. If
> > > anyone has any comments...
> >
> > Do you care to elaborate what you have tested? I.e. netmeeting with a
> > peer which is using a proxy, too, or something like this?
>
> I tested it with netmeeting. I'm not happy with it though.
> Both users in a call must be using the same gatekeeper, or
> gatekeepers which are neighbours.
>
> Users of different gatekeepers, or of no gatekeeper at all, cannot call
> you if you use a gatekeeper.
Hmm, my understanding was (untested) that it supported user@domain type
name, where the domain is looked up to find the gatekeeper. From looking at
the code, if there exists a TXT record in a domain of the form:
ras [user@]gatekeeper.host.name
then requests for user@domain will be resolved and connected using that
gatekeeper. If that fails, it will try to use the domain given as a
hostname. So user@hostname should work, no?
I don't really know how netmeeting works, but how do you usually specify the
destination? Just by ip address?
> Hence it isn't really a general purpose proxy for netmeeting.
I think I read somewhere that a truly general purpose proxy was not possible
for H323 because if an IP address is specified then the client will try to
connect directly rather than through the proxy (because the proxy pretends
to be a gatekeeper). It seems to me to be more an issue of the clients
rather than the protocol though. Any ideas?
--
Martijn van Oosterhout <kleptog@svana.org>
http://svana.org/kleptog/
Reply to: