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

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

> On 2019-11-26, at 12:29, Etienne Dysli Metref 
> <etienne.dysli-metref AT switch.ch> wrote:
> 
> Are they all necessary?
> Some namespace declarations are duplicated and in particular ns[0-6]
> doesn't make sense. Some of those are also redeclared where they are
> used. Does this serve a particular purpose?
> 
> This doesn't seem to bother our machine treatment of this feed, but my
> human brain cringes at useless "information" and duplication.

I'm inclined to agree. It looks like perhaps a union of all the specs in the 
world, perhaps with all the namespaces that anyone sends to eduGAIN. Plus 
some other stuff I find quite mysterious. Which, to be clear, shouldn't 
affect downstream processing (at least, not now that we think all the 
software which used to have limits on the number of namespace prefixes in 
scope has been retired).

I find this one a strange choice:

    xmlns:saml1md="urn:mace:shibboleth:metadata:1.0"

That namespace is not defined by or in any way related to SAML 1, but 
(originally) defined by the Shibboleth project. It would be better in my view 
to use the conventional prefix:

    xmlns:shibmd="urn:mace:shibboleth:metadata:1.0"

I also find the presence of this one interesting:

    xmlns:elab="http://eduserv.org.uk/labels";

This was one we (I) invented for UKf internal use back in the day. I am 
fairly sure we never exported that namespace to eduGAIN, but I could be 
misremembering. We removed all references to it back in 2013. It doesn't 
appear to be used in the new aggregate, and I think it would be better to 
suppress it.

Similarly:

    xmlns:ukfedlabel="http://ukfederation.org.uk/2006/11/label";

This is used internally within the UKf as a historic marker for federation 
membership and (essentially) an entity category. It isn't published to 
eduGAIN, and doesn't appear in the eduGAIN aggregate. It should be suppressed.

Similarly:

    xmlns:wayf="http://sdss.ac.uk/2006/06/WAYF";

The SDSS project was the precursor to the UK federation. The namespace was 
used as part of our pre-entity-category "hide from WAYF" mechanism. We 
stopped using this namespace in 2017. It should be suppressed.

I still have to actually do the test, but I don't expect any of the above, or 
the duplicate namespace definitions, to affect the UKf. We pretty 
aggressively discard elements and attributes in namespaces we don't support, 
and one of the last things we do before publication is to normalise what's 
left, part of which means that we end up with the namespaces we want and we 
tightly control the prefixes used on individual elements in production 
metadata. So something can come in from eduGAIN as "<aardvark:Scope>" but we 
will not publish it that way.

That's the theory, anyway. I'll confirm when I've performed the actual test.

    -- Ian

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