Working 3.1.1 IDP upgrade to 3.2.0 now broken

Cantor, Scott cantor.2 at osu.edu
Thu Dec 3 12:40:13 EST 2015


On 12/3/15, 12:36 PM, "users on behalf of Roderick Grau" <users-bounces at shibboleth.net on behalf of rgrau at albany.edu> wrote:



>Thank you Scott and Rod.  I was focused on the data connector and did not look anywhere else.
>
>Adding a sourceAttributeID to the resolver:AttributeDefinition for OPEID fixed the issue.
>
>Not having a sourceAttributeID was a working config prior to 3.2.0.

Only accidentally. It would not have been correct (not even in 2.x). You can't have a clear dependency if you don't specify what the underlying data you're dependent on is, and having it guess creates a time bomb that will break if you change the dependency to return more than it was before.

>Do you still want a bug filed?

Yes, because of the runtime exception. Those are always bugs unless the condition were out of memory or something like that. Constraint violations are just there to provide error messages for tracking down null pointer errors.

-- Scott



More information about the users mailing list