Re: [keycloak-dev] LiveOak and JavaEE libraries

On 5/16/2014 9:08 AM, Stan Silvert wrote: > On 5/16/2014 8:47 AM, Bill Burke wrote: >> On 5/16/2014 8:23 AM, Stan Silvert wrote: >>> On 5/16/2014 4:04 AM, Bolesław Dawidowicz wrote: >>>> Yes, the intention is that LiveOak doesn't require JEE container. We >>>> started with pure netty container. Currently we deploy on WildFly for >>>> several reasons, also because of KeyCloak. However we want to start >>>> stripping it down as soon as WildFly Core is available. >>> I'm working on "Keycloak without servlets" right now. The goal is to be >>> able to use Keycloak for the EAP console where servlets are verboten. >>> So far so good. >>> >>> But it begs the question, "Since the Servlet API is mostly just a facade >>> on top of the Undertow API, why do we care if Keycloak is using the >>> Servlet API?" >> There's two places we use the servlet API. In KeycloakApplication and >> in CookieHelper. The former is to get the context path of the >> deployment, the latter is because JAX-RS 1.1 API (required because of >> EAP 6.x) doesn't support HttpOnly Cookies. >> >> > Also, I see that UndertowUserSessionManagement is using the servlet's > HttpSession class. But it probably should be using Undertow's Session > class instead. > > My main point was to ask Boleslaw why it matters. I think the technical > arguments against using the Servlet API have gotten much weaker. I haven't implemented a pure Undertow adapter yet.