- From: Kingsley Idehen <kidehen@openlinksw.com>
- Date: Wed, 11 Jan 2012 08:55:16 -0500
- To: public-xg-webid@w3.org
- Message-ID: <4F0D94C4.9010202@openlinksw.com>
On 1/11/12 3:11 AM, Peter Williams wrote: > I deployed a authorization guard at my idweb site. if an OAUTH token > is sent to the webid endpoint, its evaluated using webid validation > logic, assuming it contains a cert in the OAUTH claims. If I gave out > the key, I dont see why an OAUTH v1 token could not be formualted > directly, rather than doing what is now described. > > The scenario fits one of our products, that is a thick client (using > web services, including rest-based web services). How would I make it > apply the webid world (since its not a browser)? You should be able to simulate a non browser UA using cURL re. RESTful interaction patterns. Then we get into pem, .p12, and local keystore API bindings on the client side re. WebID. > > The UA has the users signing key, shared with the browser - since they > are on the same windows machine typically. All apps on windows share > keys (given security rules). A demo UA signs a SAML2 token targetting > the my webid profile resource which the resource STS in the cloud > converts into an OAUTH v1 token (signed using an hmac known between it > and the profile's guard). This is returned to the UA, which installs > the OAUTH token into the www-auth header accompanying the web request, > which thereby forward the name and the cert (as signed by the cloud). > This only happens if the cloud has verified the certificate and SAML > assertion's signature classically (with no webid rules) The guard > receiving the www-auth header then evaluates the additional webid > rules (with no PKI rules), since it has a cert (much like SSL delivers > a cert). Failed or happy verification returns an exception (403 or 200). Yes. This is just more inter-protocol plumbing. > > Anyways got some kind of validator up in the cloud, using the cloud > STS to do the heavy validation, leaving to the webid validator > handling of the one particular extension (and modulus/exp) webid cares > about. This allows the edge devices (i.e. the cloud STS) to be > shielding mere resource servers from DOS - and enforcing proper cert > handling, cert chain discovery etc. > > Not going anywhere, but its fun to see how webid will surely evolve - > and work with all the othe security technology families. Seems to have > a nice fit, once divorced from SSL. > > Ok. tomorrow, can try again at SSL. Lets finally see what Azure does > with client certs, since there is load balancer in the way. Okay. -- Regards, Kingsley Idehen Founder& CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca handle: @kidehen Google+ Profile: https://plus.google.com/112399767740508618350/about LinkedIn Profile: http://www.linkedin.com/in/kidehen
Attachments
- application/pkcs7-signature attachment: S/MIME Cryptographic Signature
Received on Wednesday, 11 January 2012 13:55:53 UTC