[OpenID] why is xri so obtuse?

Martin Atkins mart at degeneration.co.uk
Tue Jan 2 01:18:06 UTC 2007

Greg Hewgill wrote:
> Now, in order to try to catch up with all the latest developments, I've
> tried to read about XRI and i-names and now I'm completely baffled.

When I raised a similar concern about XRI, I was asked to be a bit more 
specific about what was bothering me. Since then, I have been making an 
effort to try to figure all of this stuff out, and these are the main 
obstacles I have hit:

  * The information about all this stuff is spread all over the shop and 
it's near-impossible to find answers to specific questions.

  * The specs all suffer from "out-of-scope-itis": that is, every 
document is an incomplete picture and references a bunch of other 
documents, each of which themselves reference still more documents. 
There doesn't seem to be anywhere to get the "big picture" or a 
description of how this architecture operates in the common cases.

  * As a specific instance of the above, I've been unable to find 
anywhere that tells me how to resolve XRIs under the GCS roots. (For the 
uninitiated, these are XRIs that start with =, @, +, $ or !). The 
closest I've come is the XRI 2.0 FAQ[1], which states:

     Resolution of an XRI rooted by a GCS character requires
     pre-configuration of a resolver (or its proxy) with the location
     of the GCS root. This is out of scope for the XRI specification

   I understand the need to compartmentalize the specs, but it seems 
ridiculous that this particular information is so hard to find given 
that you absolutely can't resolve most XRIs without it!

  * The existing implementations out there seem to be largely out of 
date, as far as I can tell. The Perl module "XRI" on CPAN doesn't work 
because (I assume) it has outdated configuration for the GCS roots. I 
guess the Java OpenXRI implementation is up-to-date, but as is typical 
of Java libraries there appears to be no simple example or "common case" 
easy API, so you already have to understand nearly everything about XRI 
before you can begin to grok the library.

   Ideally I'd just like a simple tool like the xriresolve.pl that comes 
with the CPAN XRI package which does resolution of an arbitrary XRI 
while printing out verbose progress information so that I can understand 
what it's doing. If someone would point me to an up-to-date one of these 
I'd be very grateful.


My main suggestion to the XRI folks would be to put up some introductory 
documents explaining the technical aspects of XRI and some complete 
examples of transactions for common tasks like resolving an XRI to an 
XRDS (both with and without a proxy resolver) or interacting with the 
forwarding service. If you've already got such documents, please both 
point me at them and publish links to them in a more obvious place!

[1] http://www.oasis-open.org/committees/xri/faq.php#_Toc121168695

More information about the general mailing list