Shibboleth Developers

[Ian Young <>]

Scott Cantor wrote:

> I wasn't aware any were missing, but it's certainly possible.

Quick grep|wc results:

IQ has 266 entities (191 IdPs, 77 SPs and yes that doesn't add up 
because some have both roles) and only 157 NameIDFormats.

InCommon has 28 entities (14 of each kind) and 14 NameIDFormats.  That 
14 is a coincidence; there are both IdPs and SPs with and without a 
NameIDFormat.

> I thought your
> style sheet created them. If not, that would explain why they aren't there.

No, my 1.2 -> 1.3 converter doesn't seem to include these.  Turns out we 
discussed this in mid-March and eventually decided not to, because:

> Shib requires that implementations support that format, but it doesn't
> require anybody deploy it. Metadata reflects deployment, not implementation,
> so the values depend entirely on the deployment choices.

Maybe I misinterpreted that?

> I thought we had, but since InQueue is soon to die, I'm not about to worry
> about it now.

Seems to apply to InCommon, too, FWIW, from the figures above. 
Obviously not scary, but more of an enduring anomaly than InQueue.

> I can't see any way it would break something, but if somebody were altering
> their deployment to use some other Format, certainly their metadata ought to
> reflect that choice.

Meanwhile, this statement is pretty comforting and definitely supports 
the idea that we should have included this in the conversion in the 
first place.  Sorry if I fluffed that.

        -- Ian