[OpenID] Interop

[Guido Sohne]

On 6/8/07, Martin Atkins <mart at degeneration.co.uk> wrote:
> I don't know if this was what Fen was referring to, but the XRI syntax
> *is* actually heirarchical, so you can create your own namespace and
> still be an XRI.

I understand this.

> There are a number of approaches. You can register an organisational
> i-name and then put other people under it:
>      @myorganisation*fred

Ok. Where? We are running into a centralization/p2p issue here. Some
people are going to want verification and trust services that emerge
from a centrally administered directory and other people are going to
want to run their own.

Implementing XRI (I haven't looked at it in any depth at all) appears
to be akin to maintaining a database (a huge database!) and I think
this is what you are referring to by  that it is hierarchical, so an
organization could be in control of that segment of the 'global' data?

> You can also embed your DNS hostname directly into the XRI to avoid the
> need to have an i-name at all. I can't remember the exact syntax,
> unfortunately.

It looks more and more like XRI could be a replacement for DNS itself?
If DNS information were subsumed into the XRI databases, you could
simply toss the DNS system in the bin ...

And the same problems are cropping up again. I believe what we are
discussing is akin to the multiple roots issues (regular DNS versus
OpenDNS) seems strikingly similar to (xri.net versus 'some other'
provider).

> The "funny little symbols" (=, @, etc) are really just syntactic sugar
> for some nominated "special" prefixes.

Nice to have. But who will use it? The record of people using search
engine operators for example, or the difficulty with typing / grokking
URLs doesn't make me feel very optimistic. But is is certainly better
than hugely long URLs based on conventions.

And is gets even clearer that OpenID should decide between XRIs and
URLs IMHO. The time to avoid future messes would be ... now!

-- G.