Sounds like we need some test cases for this - it appears that the exception you're seeing is if there is no Subject accessible to the thread. That shouldn't be the case since the ShiroFilter should have bound the subject to the thread via the WebSubjectBuilder and WebThreadStateManager usage.
On Mon, Aug 24, 2009 at 1:56 AM, Kalle Korhonen<[email protected]> wrote: > No.. this can't be right - for example calling > SecurityUtils.getSubject().login(authenticationToken) results in: > java.lang.IllegalStateException: Subject context map must contain a > javax.servlet.ServletRequest instance to support Web Subject > construction. DefaultSecurityManager's createSubject operations create > Subject context map from scratch, and obviously it won't have the > required objects in the context. Les, care to clarify your refactoring > plan and how this is supposed to work? > > Kalle > > > On Sat, Aug 22, 2009 at 11:22 PM, Kalle > Korhonen<[email protected]> wrote: >> I see the internals of Shiro have been changed quite a bit in r806735. >> ShiroFilter.bind() now does: >> Subject subject = new WebSubjectBuilder(getSecurityManager(), >> request, response).build(); >> WebThreadStateManager threadState = new >> WebThreadStateManager(subject, request, response); >> threadState.bindThreadState(); >> >> which for Tapestry integration I'm working on results in: >> >> java.lang.IllegalStateException: No ServletRequest found in >> ThreadContext. Make sure WebUtils.bind() is being called. (typically >> called by ShiroFilter) This could also happen when running >> integration tests that don't properly call WebUtils.bind(). >> at >> org.apache.shiro.web.WebUtils.getRequiredServletRequest(WebUtils.java:351) >> at >> org.apache.shiro.web.session.ServletContainerSessionManager.doGetSession(ServletContainerSessionManager.java:69) >> at >> org.apache.shiro.session.mgt.AbstractSessionManager.getSession(AbstractSessionManager.java:246) >> at >> org.apache.shiro.session.mgt.AbstractSessionManager.checkValid(AbstractSessionManager.java:265) >> at >> org.apache.shiro.mgt.SessionsSecurityManager.checkValid(SessionsSecurityManager.java:294) >> at >> org.apache.shiro.mgt.DefaultSecurityManager.getSession(DefaultSecurityManager.java:196) >> at >> org.apache.shiro.mgt.DefaultSecurityManager.resolveSessionIfNecessary(DefaultSecurityManager.java:437) >> at >> org.apache.shiro.mgt.DefaultSecurityManager.getSubject(DefaultSecurityManager.java:403) >> at >> org.apache.shiro.subject.SubjectBuilder.build(SubjectBuilder.java:95) >> at >> org.trailsframework.security.services.SecurityConfiguration.service(SecurityConfiguration.java:87) >> >> I.e. WebSubject requires the request is already bound to thread >> context, but WebThreadStateManager (that's supposed to bind it) >> requires a subject to exist. If I call WebUtils.bind(request) >> before instantiating a WebSubjectBuilder, everything works. Les, is it >> expected I still need to bind the request/response separately or >> perhaps this is a defect/refactoring still in progress? >> >> Kalle >> >
