NameIds (again)

[Cantor, Scott]

On 9/29/16, 8:25 AM, "users on behalf of Dave Perry" <users-bounces at shibboleth.net on behalf of Dave.Perry at hull-college.ac.uk> wrote:

> I fixed it by adding an encoder (I thought someone had said you didn't need to, if an
> attribute was purely for NameID generation)

That is correct.
 
> and setting the encoder type as Scoped string vs NameID.

I don't know what that refers to. Using a deprecated StringNameID encoder would work but is just the alternative means of generating a NameID. Still works certainly, I use that myself due to my config being older/upgraded.

But adding just any old encoder to an AttributeDefinition would not have any impact on whether the resolver surfaces the attribute or not. At least I'm not aware that it does, that would be a bug.

Was the definition marked "internal"? If not, if removing the encoder causes the attribute to disappear internally, there's a bug.

-- Scott