I found it too. What I did was trying to not use getId() when cleaning up
that session. I have put all needed info into the hashtable within that
session so I still be able to get it out via "this.xxx" object.
Steve N.
E-mail: [EMAIL PROTECTED]
----------------------------------------------
Presents KBMail Professional Service
Visit www.kbmail.com
Access to your existing POP3/IMAP4 via Web browser for FREE
----- Original Message -----
From: Brett Peterson <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Saturday, April 24, 1999 6:26 AM
Subject: Possible HttpSession.invalidate() bug
> I have an issue with JServ's current implementation (1.0b3) of the
> HttpSession.invalidate() method. I believe what I found is a bug, but
that
> is debatable. I'd appreciate any comments or pointers to where this has
> been discussed previously.
>
> Here's the situation:
>
> I am piggy-backing on top of JServ's session management thread in order to
> clean up an application-specific state table. Specifically, I am using
the
> HttpSessionBindingListener interface to get a callback when I bind or
unbind
> an object to a session. If you're not familiar with this interface, it
> contains the following two methods:
>
> public void valueBound(HttpSessionBindingEvent event);
> public void valueUnbound(HttpSessionBindingEvent event);
>
> The HttpSessionBindingEvent class (the parameter to both calls) has a
method
> with the following signature:
>
> public HttpSession getSession();
>
> This method returns the session that the object is being bound to (in the
> case of valueBound()) or is being unbound from (in the case of
> valueUnbound()).
>
> My first design/implementation of valueUnbound() required that I grab the
> session from the event and then call the getId() method on the HttpSession
> object. This is where the trouble started.
>
> When a session is invalidated in JServ, the "valid" flag within the
> HttpSession object is immediately set to false. After that, all
> HttpSessionBindingListener objects that were bound to the session have
their
> valueUnbound(event) method called. Because the session was first
> invalidated (by setting the valid flag to false), any method calls on the
> session will throw an IllegalStateException. Since my valueUnbound()
method
> did make a call on the session (i.e., getId()), the exception was thrown.
>
> This seems wrong to me. The invalidate() method should call all the
> valueUnbound() methods first and then set the valid flag to false. The
way
> it is now, there is no reason to pass the session in as part of the event
> because you cannot query the session for any useful information (all
methods
> except isValid() will throw an IllegalStateException).
>
> The JServ implementation of HttpSession may be found here (my email client
> will split the URL). Take a look at the invalidate() method:
>
>
http://www.working-dogs.com/cvsweb/index.cgi/~checkout~/jserv/src/java/org/a
> pache/jserv/JServSession.java?rev=1.5&content-type=text/plain
>
> Is the current implementation of HttpSession.invalidate() correct or is my
> suggested alteration correct? I was able to get around the issue with
some
> refactoring, but the resulting code was not as elegant.
>
> Thanks,
> Brett Peterson
> [EMAIL PROTECTED]
>
>
>
> ----------------------------------------------------------------
> To subscribe: [EMAIL PROTECTED]
> To unsubscribe: [EMAIL PROTECTED]
> Archives and Other: <http://java.apache.org/main/mail.html/>
> Problems?: [EMAIL PROTECTED]
>
----------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Archives and Other: <http://java.apache.org/main/mail.html/>
Problems?: [EMAIL PROTECTED]
