what controls assertion signing?

[Mark K. Miller]

On Fri, 6 Feb 2015, Cantor, Scott wrote:

>> If IDP config doesn't trump what others put into metadata you shoulf
>> file a bug, IMO.
>
> It does, in fact. We didn't see it as a bug because metadata is never 
> viewed as something somebody can accidentally inject bad data into, it's 
> always implicitly trusted.

You do know metadata comes straight from vendors out here in the wild when 
we're just testing, right?  ;-)

Seriously, though, in the controlled federation scenario, I'd have no need 
at all to argue with that explanation.

>                            Apart from performance, there's no real 
> impact to signing them.
>
> We can argue about what "never" should have meant, but that's what Chad 
> defined it to mean ("never, except when metadata says to").
>
> That said, I think I completely forgot to implement that flag in V3. So 
> my inclination right now is to just define a property for whether to 
> honor it, and default it true to match V2 behavior.

You're welcome!  ;-)

> -- Scott
>
> -- 
> To unsubscribe from this list send an email to users-unsubscribe at shibboleth.net