On 13/09/2010, at 8:42 PM, Nicolas Williams wrote:
> On Mon, Sep 13, 2010 at 08:41:29PM +0200, Luke Howard wrote:
>>> Well, I _had_ proposed a standard interface for them and then agreed
>>> with Leif to drop it. However, if people really would use a map-to-any
>>> API, and then I think standardizing it is not a bad idea -- I just would
>>> want to see that support before I try again.
>>
>> I think the difference here is that both SAML and local attributes can
>> be surfaced via gss_get_name_attribute(), because they are, erm, just
>> attributes.
>
> Yes, but not as native types -- attribute values are octet strings.
Surfacing both SAML and local attributes as octet strings is still useful in itself. Surfacing them as native Shibboleth and/or OpenSAML objects may also be useful to some applications, but that is (more or less) orthogonal.
-- Luke
|