You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@wicket.apache.org by "Torsten Krah (JIRA)" <ji...@apache.org> on 2017/11/10 10:46:00 UTC

[jira] [Created] (WICKET-6493) WebSocket SessionIds are wrong (HttpSession one used instead of Websocket one) + NPE if no HttpSession is found during Handshake Request

Torsten Krah created WICKET-6493:
------------------------------------

             Summary: WebSocket SessionIds are wrong (HttpSession one used instead of Websocket one) + NPE if no HttpSession is found during Handshake Request
                 Key: WICKET-6493
                 URL: https://issues.apache.org/jira/browse/WICKET-6493
             Project: Wicket
          Issue Type: Bug
          Components: wicket-native-websocket
    Affects Versions: 7.7.0
         Environment: Linux, JDK 8, Jetty 9
            Reporter: Torsten Krah
            Assignee: Martin Grigorov


Reading the *HandshakeRequest* javadoc this is written:

{code}
    /**
     * Return a reference to the HttpSession that the web socket handshake that 
     * started this conversation was part of, if the implementation
     * is part of a Java EE web container.
     *
     * @return the http session or {@code null} if either the websocket
     * implementation is not part of a Java EE web container, or there is
     * no HttpSession associated with the opening handshake request.
     */
    Object getHttpSession();
{code}

So its valid to have *NO* *HttpSession* assigned to a JSR356 request.

But Wicket seems to "expect" this because in *JavaxUpgradeHttpRequest* the HttpSession is extracted from the *userProperties* which is not there, if it was "null" in the HandshakeRequest.
In the processing the chain will reach the *AbstractWebSocketProcessor* constructor which calls

{code}
this.sessionId = request.getSession(true).getId();
{code}

which is going to fail here because there is no HttpSession on the crafted request used here.

I wonder why the *sessionId* is taken from the HttpSession at all - shouldn't it be the JSR356 session id which you can get with *session.getId()*.
If you look e.g. at the *IWebSocketConnectionRegistry* javadocs most methods have their *sessionId* argument documented with:

{code}
	 * @param sessionId
	 *      the web socket client session id
{code}

but actually the HttpSession id is used instead of the WebSocket sessionId - which is a totally different one.

So i am proposing to use the "real" websocket session id here taken from the constructor call:

{code}
public JavaxWebSocketProcessor(JavaxUpgradeHttpRequest request, final Session session, final WebApplication application, EndpointConfig endpointConfig)
{code}

where there is the *session* argument where *session.getId()* actually returns the websocket session id which can last longer and is 100% there - the HttpSession must not be there at all as written.

Unfortunately the javadoc on *IWebSocketConnectionRegistry* for

{code}
	/**
	 * @param application
	 *            the web application to look in
	 * @param sessionId
	 * @return collection of web socket connection used by a client with the given session id
	 */
	Collection<IWebSocketConnection> getConnections(Application application, String sessionId);
{code}

does not state which "sessionId" is meant here - i guess its the *HttpSession* - in that case please clarify the JavaDocs here.
The code maybe need to "track"the HttpSession id to handle that usecase - but in an optional way because it may not be there.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)