An open discussion list for topics related to the eduGAIN interfederation service.

[Ian Young <ian AT iay.org.uk>]

On 7 Jul 2014, at 11:30, Glenn Wearen <glenn.wearen AT heanet.ie> wrote:

>> So please take away italian entities from the ukfederation metadata.
> 
>> We can do that if you request it, but it would mean that no UKf entities 
>> AT ALL, whether opted in to eduGAIN or not, will see imported entities 
>> from your federation. That doesn't sound to me like the best solution.
> 
> Could you remove the Irish institutions from the UKf wayf but leave them in 
> your metadata? I know there’ll still be problems with SP’s who operate 
> their own DS, but the UKf must account for half of all discoveries?
> 
> I don’t mind reaching out UKf SP’s to remove Irish institutions from their 
> DS’s, I’ll also have to reach out to SP’s who operate in eduGAIN that are 
> listing Irish institutions under the UK federation like the Terena SP.

It would be possible for us to hide selected IdPs from our default CDS UI, 
but have them come up if the "search across all sites" option was selected. 
We did consider that originally, but felt that exposing all eduGAIN IdPs by 
default was more likely to be what people would want. I do remember us 
concluding when it was originally discussed internally that this would be 
something we might reconsider, so it's not carved in stone, but I think we'd 
be a little nervous about changing something that has been in place for so 
long. We'd have to consult fairly widely internally for a change like that 
for all eduGAIN IdPs.

By saying "selected IdPs" above, though, I'm saying that if people wanted us 
to "hide from DS" on a per-registrar basis then that would be possible in 
principle. We don't have mechanisms in place to allow that today, but we 
could develop them if there was agreement that we could improve things for 
everyone by going down that route and giving you more choice in the matter. 
It wouldn't involve a change in our metadata republication policy and would 
be the responsibility of the individual registrar.

I'd also emphasise that such a change wouldn't help at all with Lalla's 
specific problem, as that is with an SP which does not use our CDS anyway, 
but instead uses a (very inclusive, try scrolling to the end of the list) 
local DS of their own devising.

        -- Ian

Attachment: smime.p7s
Description: S/MIME cryptographic signature