[OpenID] Making sure OpenID works (was: Starting new OpenID Workgroups)

[SitG Admin]

>Traditionally, what people have done is defined interoperability 
>test suites, required that all vendors pass the test suites, and 
>made the use of the brand (here: the OpenID logo) dependent on 
>passing the test suites.
>
>So far, we have not done this in OpenID.
>
>Do we need to?

I think test suites (assuming, here, that you mean "3rd-party sites 
will try to interact with your implementation and diagnose the 
problem") help with people who are already working on it, but we 
should also look at assisting beginning developers with understanding 
the process - not just those who already have reason to work on it 
(because they're interested, or they've been paid), but those who are 
trying to figure out how OpenID *works*. Increased transparency in at 
least one library, even an example library constructed specifically 
for that purpose (as an "ideal flow" for OpenID), would help there. 
Instead of just saying "We're open-source, look at it!" (which 
requires some level of time and thought investment into reading and 
understanding our source) and pointing at the specs and some 
higher-level flow descriptions, let's bring developers as close as we 
can to the actual operation: the equivalent of setting PHP's 
error-reporting level to ALL, where they constantly see information 
printed about each stage of the process they're on and what's going 
on overall (where variables are having their values stored for later 
use, where they can be effectively discarded).

-Shade