Karsten Otto wrote:

> Ok, I see. But this implies there could be more servers than just  
> one, each hosting a replica. Which one do I contact for updates? With  
> VOP/VIP URLs this was straightforward, but please remind me again,  
> how do I contact a vos:0011223344... key-based site? Is there a name  
> server somewhere?

There are two layers, host and site, where before there was just a site.

Peter, does the following sound more or less like the way it's going to 
work or could work, roughly speaking? --   First, you contact a host, 
then you find the site and its vobjects.  You either already have a 
connection, or you have a URL like 
vip://interreality.org:4231/MyCoolWorld/thing.   You could say that 
"interreality.org:4231" is the name of the host.   Then you go looking 
for the /MyCoolWorld/thing vobject, which is part of some site hosted on 
this host.  You could just send the host the name of your vobject 
("/MyCoolWorld/thing), and the remote host looks up which of its sites 
(if it has more than one) that vobject is part of [vobjects on different 
sites would have to be children of different "root vobjects", like 
/site1, /site2, /MyCoolWorld], and returns a message informing you of 
your requested vobject's site identity information, you do key 
authentication, etc.  The site doesn't really have a name, but it has 
its key, and on any given host it has a unique root vobject that all its 
other vobjects are underneath (though that root could have different 
names on different hosts).    This is not too dissimilar from the site 
handshaking/negotiation in S4, I think, it just has the extra part where 
you need to tell the remote site what vobject you want so it can figure 
out which site to use for that vobject, and tell you.

Yes, nobody should really have to every care about the actual site 
ID/key strings, unless they're debugging something or it happens to be 
useful in some code to have a unique value that  identifies the site 
(i.e. in S4 you would have used the site's URL for that) or are 
developing something more advanced involving site replication or 
distribution or whatever.  Or they are suspicious about a site's 
identity and want to manually verify whether it matches some record . 
End users ("surfers") would still have URLs.

Though one thing I do want to get into the client application etc. more 
quickly than we would have in S4, is reimplementing the service 
directory we had in S3.  It incorporated both online directories of 
"services" (e.g. a 3D world), and LAN discovery.


vos-d mailing list

Reply via email to