On Fri, Oct 15, 2010 at 5:52 AM, Moritz Gmelin wrote:
> Hi,
>
> Howard was right. It was a race condition.
> Maybe for other to prevent similar problems:
>
> We have an AccessControlDispatcher in our application that was inserted
> before:ComponentEvent in our application. This AccessControlDispa
Hi,
Howard was right. It was a race condition.
Maybe for other to prevent similar problems:
We have an AccessControlDispatcher in our application that was inserted
before:ComponentEvent in our application. This AccessControlDispatcher uses the
SessionStateObject to check access permissions.
N
I don't ever see Jetty starting strangely ... what else is going on in
your application besides Tapestry.
Also ... it's always a good idea to "audit" your runtime classpath,
make sure there isn't anything odd going on there.
On Thu, Oct 14, 2010 at 2:17 PM, Moritz Gmelin wrote:
> Getting closer.
Getting closer.
In the cases where it happens, the binding and unbinding is called after
cleanup from different threads. Threads are named btpool0-1 -btpool0-7.
But as I said, this happens only about every second time I start my jetty
server. The other times, the binding- unbinding is called on
Hi,
I added the logging as you suggested. This is part of the logging for a
_SINGLE_ page request. CurrentSessionASO is our SSO.
10-10-14 21:51:12.049 [DEBUG]
de.example.aso.CurrentSessionASO.valueBound(CurrentSessionASO.java:1587)
session 1vsqjhkzdsj37 asoHash 1363402032
10-10-14 21:
Tapestry manages persistent state in the HttpSession. As of 5.2, it
doesn't do anything special to handle cases of multiple requests
(perhaps due to Ajax) attempting to modify the same SSO in multiple
threads.
This aligns with Brian Goetz's observation that all stateful web
applications are inher
Hi,
since upgradeing to Tapestry 5.2 (5.2.0 and 5.2.1) we sometimes obseve that the
contents of our SessionStateObject gets lost.
The Session-ID between 2 requests is identical, but the Hashcode of the
@SessionState object changes and thus the contents are lost.
If everything works correctly, t
