Hi,

Can you submit some evidence that shows that URIs should be normalized
before comparison?

Colm.

On Fri, Jun 5, 2020 at 6:33 PM Nimish Telang <[email protected]> wrote:

> Hi,
>
>
>
> We have encountered an issue where https://example.com and
> https://example.com:443 don’t match when included in SAML audience
> restrictions.
>
>
>
> As far as I can tell this is because the code to validate matching is just
> basic string comparison (And should probably use Sets rather than Lists):
> https://github.com/apache/ws-wss4j/blob/74df6178e87edbf28b267845c7dcaa5203df5eca/ws-security-common/src/main/java/org/apache/wss4j/common/saml/SamlAssertionWrapper.java#L902
>
>
>
> Using java.net.URI to resolve this would be handy – also as the actual
> type that should be parsed from the SAML, not String, since the type in the
> xsd is anyURI…
>
>
>
>
>
> I don’t see a security issue here, and the specs for the saml core
> recommend doing it for at least the authz stuff.
>
>
>
> Nimish
>

Reply via email to