[ticker-dev] Re: Elvin presence spec draft - request for comments

[Phillips, Matthew]

Hi all,

This is a kind of consolidated reply to Klaus, Bill, Martin and David's
comments.

Klaus: your interpretation of how the RFC relates to Sticker's
implementation is spot on.

All: I guess I'm actually proposing a spec-within-a-spec for the presence
RFC, namely that we adopt a unique user naming scheme for ticker similar to
email, where local usernames are resolved using logical domains.  Right now,
people are using "name at machine-name", "person at home", "person at dstc", etc as
de-facto names, mainly to communicate not only who they are, but where they
are at the moment.  This can be confusing, not least of all because I need
to know that bill at home and bill at dstc are the same person, but (probably)
bill at large is not.  So what I'm suggesting is that we formalise the thing:
pick a domain that suits you best and stick with it.  Other than its role as
a username resolution device, the domain can be whatever you like (although
using your organisation or DNS domain would be logical).  As Bill suggests,
other information, such as where you are, can go into the presence info.

I really, really want to avoid having a binary crud unique 'user id'.

The other reason a domain is useful is that it gives you a useful starting
buddy list that is automatically populated and maintained (note that being
able to see everyone in the whole of tickerland is unscalable).  When you
(inevitably) want to add people outside your domain to your buddy list, your
client should make it easy to do this.  So Klaus, you've hit a problem in
the Sticker UI, not in the spec - you can't see the people you want to
because you can't add them to your buddy list.  A way of querying what
domains there are are would be good too, but I suspect this would probably
require a central server that is subscribed to all Presence-Request's.  

I'll have another look at IMPP and Jabber to see if there is any wisdom to
be found there.

Cheers,

Matthew.

> -----Original Message-----
> From: kas-list at magnetic-ink.dk [mailto:kas-list at magnetic-ink.dk]
> Sent: Thursday, 27 September 2001 12:36 AM
> To: Tickertape Development
> Subject: [ticker-dev] Re: Elvin presence spec draft - request for
> comments
> 
> 
>   G'day mates,
> 
> With keen interest I've been flipping through Matthew Phillips' neat
> PowerPoint slides, "Awareness in Tickerland", from the Ticker2001
> workshop.
> 
> I've also been snooping a bit on Sticker 2's ELVIN_ONLINE_STATUS*
> emissions and was just about to implement something similar in my
> roll-your-own tickertape client written in python when I stumbled
> upon the Elvin Presence RFC on the web.
> 
> I assume that the ELVIN_ONLINE_STATUS_REQUEST from the slides equals
> the Presence-Request from the RFC, and that the ELVIN_ONLINE_STATUS
> corresponds to the Presence-Info -- in spirit, not in word -- and I
> have a question about the implementation.
> 
> The way I read it -- and feel free to correct me -- the presence
> protocol would not be very useful for me the way Tickerland looks
> these days: there's a bunch of guys (and a few gals) Down Under, and
> then there's me and my shadow at the other end of the world.  When
> I start my ticker client it is interesting for me to know who's
> online in Australia since they are my only company in Tickerland
> (oh, the other day a Danish company, probably mistakenly, popped
> up on my tickertape and said something about a fax or a printer,
> but I haven't seen them since...).
> 
> Now, Sticker2 seems to make two subscriptions [for me] on startup:
> 
> <pre>
> require(ELVIN_ONLINE_STATUS) && DOMAIN == "zigzag" && USER != "kas"
> require(ELVIN_ONLINE_STATUS_REQUEST) && DOMAIN == "zigzag"
> </pre>
> 
> i.e., it asks for presence info from anybody in my own domain, and
> since I'm the only one in my domain who uses Elvin messaging (well,
> probably in Denmark, or even the entire Europe) there will never be
> any answers to the status requests, will there? :-(
> 
> I suppose I could make a different implementation in my own ticker
> client, but I still see it as a problem  -- and now I'm getting to
> the point -- that the proposed presence protocol is domain focused.
> 
> I am usually elvining as kas at zigzag.  Then there's j at laptop (or
> @lerkim or @monash, or whatever).  And d at monash, etc., etc.  How,
> and where, am I gonna catch those boys if my client only wants to
> subscribe to presence info from my own lonesome domain?
> 
> Perhaps it's beyond the scope of the presence protocol, and so
> perhaps I should rather be using ICQ for this, but when I start up
> my ticker I would like to be able to see who's online "globally" --
> that e.g. Julian, Bill, David and Martin are online Down Under,
> even if they're on separate domains, doing different things.
> I admit that if Elvin ever gets as accepted worldwide as ICQ is
> now, then we'll end up with an amazing buddy list, but I am
> looking at the situation and my requirements here and now.
> 
> Assuming clients are ~/.ticker/ aware, would it make sense to have,
> e.g., a ~/.ticker/buddies file, perhaps along the lines of
> 
> <pre>
> # Name		Usually seen as
> Julian		j@(laptop|lerkim|monash)
> Christopher	cmlp@(dstc|home)
> # eof
> </pre>
> 
> and then have your client make specific presence requests for all
> user at domain entities mentioned in that file?  Or is there a more
> efficient way to accomplish what I want?
> 
> Or have I misunderstood the whole thing?
> 
> 
> Cheers,
> 
>   // Klaus
> 
> -- 
> A man, just one -
> also a fly, just one -
> in the huge drawing room.
>