[sldev] Question: Replacing current group chat with XMPP?

[David M Chess]

From: "Christian Scholz / Tao Takashi (SL)" <tao.takashi at googlemail.com>

> I also wonder if the group situation will stay that way on the long
> run with OGP. Right now with only 25 groups you probably need to stay
> in the big groups to stay connected. If you wouldn't have that
> requirment and the whole world is more decentralized anyway I wonder
> if those groups would look different. Additionally if you are more
> free to join and leave without any hassle (like you might do with IRC
> channels). Right now with that limit you always have to think plus I
> think some groups have member restrictions because of the lag (IIRC).

Just as a datapoint, I as a private Resident would be quite unhappy with a 
change to the group system such that I had to explicitly "join" the Group 
IM channel for all my chat groups once per logon (say) in order to see the 
Group IM. 

> But of course we also need to find a solution to support the existing
> group model with those huge groups.
> BTW, what are the actual numbers for the bigger ones?

Some random data points: Money Island VIP (raffle) 9984, Fashion 
Consolidated (open) 9009, The Wild Coast (pay to join) 5640, Eightbar 
(invitation only) 4600, Lucky Money Chair (open) 4529, Scripters of Second 
Life (geeks lol) 1974.

> So pluggability might be key and the question is how we can support
> both: Some not so secure environment where you maybe can simply join
> some non-SL IM system like Jabber or IRC without further identity
> check 

Presumably that can be done by just embedding an IRC or Jabber client into 
some viewer.  Or just running it in a different process on the same 
machine.  :)  No OGP implications, I don't think.  I don't see a strong 
use-case for it, myself; why should an SL viewer contain a way for me to 
chat with random people whos RL and SL identities I don't know?  We 
already have programs for that.

> and a more secure one like the one we have now where you can be
> sure that "Tao Takashi at syntronik.de" (imagine this would be my full
> qualified name in OGP) really is me. Additionally there might be a mix
> of both but where you can distinguish not verified agents as such.

The secure one seems like the primary use-case to me; the mixed case would 
be a nice addition, where people can come into the secure system without 
being verified, as long as they're clearly marked as such.

> Also I think the IM service should be kept outside the AD but can ask
> the AD for verification of somebody.

That's roughly what the current proposal on the Wiki says: use the AD for 
identity, but don't flow the actual message traffic through it.  Your 
comments there would be most welcome, although I know you hate the poor 
Wiki.  :)

> So that means you have to trust the IM service to do the
> authentication and of course the ADs in question to answer it
> correctly. The details "just" need to be worked out ;-)

My preference would be to do this by having the IM service under the 
control of the AD; avoids having another entity to have to trust.

> Another thing I am thinking about is how we can also use service
> discovery (e.g. by using XRDS-Simple or the successor protocol which
> is worked on right now) for each person to check which IM service it
> prefers to be contacted under.

Would seem at least as logical to me for the AD to know that (since the AD 
knows other stuff about me anyway).  But whatever works.  :)

Dale Innis
DaleInnisEmail at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.secondlife.com/pipermail/sldev/attachments/20080908/c054e527/attachment.htm