Killing time

[Carl Howells]

Brad Fitzpatrick wrote:
> Here's a draft:
> 
>    http://openid.net/specs-notime.bml
> 
> Once I get your approval, and nobody else objects, I'll copy that to the
> real specs.bml page.

A couple more things...

First, it probably should be clarified that replace_after_s (or 
expires_in, if it's renamed) is an integer represented in human-readable 
base 10.  That's probably the obvious thing to assume, but it's better 
to make it explicit.

Second, I'd change the wording for the note on that field, regardless of 
what it's named, to something like this:

"The server should not honor this association after this number of 
seconds has passed.  The consumer should create a new association at 
that point instead of attempting to re-use this one."

The use of "should not" is a bit awkward as it's RFC-like language, but 
I think it's closest to the intended meaning without sounding as vague 
as "probably won't", which was my second choice for the wording.

Carl