Backwards compatibility

[Brad Fitzpatrick]

On Mon, 25 Sep 2006, Dick Hardt wrote:

> Given there is so little difference between 1.1 and 2.0, and one of
> them being support for extensions, I am confused
> why you would not just support 2.0.

You can do extensions with 1.1 too.  It's just not really described well
enough.

But let me make it more clear:

The point of LiveJournal using 1.1 is as a forcing function to make sure
2.0 implementations obey the "MUST work with 1.1" part of the 2.0 spec.

Now, if we had a good test suite for validating implementations, then I'd
have less resistance to putting LiveJournal on 2.0.  Until that happens
(and I've love to help or lead with that part), I'm going to keep
LiveJournal as the de-facto 1.1 test suite, lame of a solution as it is.

> I was looking to clarify names for people. Not a huge issue, what became
> the issue was how much compatability was needed.
>
> You seemed to like renaming trust_root to realm! :-)

I do.  I should've called it realm to start with, and it's a confusing
enough issue to be worth changing.  Also because it's easy to document in
the spec:

   "previous versions of OpenID called this parameter 'trust_root'.
    For compatibility, you MUST treat 'trust_root' and 'realm' as
    identical.  If both are present, you SHOULD use the value in 'realm'."

... or something.

Whereas changing message formats is a lot more verbose of a change when it
comes to describing old vs. new specs.

Anyway ... yeah, moving forward ... is anybody working on a test suite?

- Brad