How would the client send the data to a broker if it were transient? :)
On 7/19/06, Hiram Chirino <[EMAIL PROTECTED]> wrote:
Adding a transient field to the ConnectionInfo would not cause a wireformat change. I'd rather keep it simple and do it in our current one. On 7/19/06, James Strachan <[EMAIL PROTECTED]> wrote: > > How about to avoid breaking backwards compatibiility (or introducing a > new OpenWire version) just adding a new derivation of a ConnectionInfo > which could take additional fields like the certificate & token etc? > > On 7/19/06, Hiram Chirino <[EMAIL PROTECTED]> wrote: > > Hi Sepand! > > > > On 7/18/06, Sepand M <[EMAIL PROTECTED]> wrote: > > > > > > Thanks for the info James. > > > > > > Please tell me what you think of the following: > > > I plan on having a new transport class that will do SSL client > > > certificate authentication and then override the ConnectionInfo > > > class's username field (we don't need a password) with the > > > distinguished name of the client. > > > > > > > > This sounds good. > > > > Now, that I think of it. I think that would useful for the > authorization > > layer to be able to access the whole cert that was provided that the > > transport layer. Would it help if the ConnectionInfo class had an > > additional transient Object field that you could attach certificate info > to? > > > > I could have the transport override every command's username with the > > > DN, but that's not needed if I use a UserIdBroker. > > > > > > That sounds good too. > > > > This should be ok, right? > > > > > > Thanks, > > > Sepand > > > > > > > > > > > -- > > Regards, > > Hiram > > > > Blog: http://hiramchirino.com > > > > > > > -- > > James > ------- > http://radio.weblogs.com/0112098/ > -- Regards, Hiram Blog: http://hiramchirino.com
-- James ------- http://radio.weblogs.com/0112098/
