DO NOT REPLY [Bug 22513] New: - Scope "leaks" using multiple sessions using new flow framework

[bu...@apache.org]

DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG 
RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT
<http://nagoya.apache.org/bugzilla/show_bug.cgi?id=22513>.
ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND 
INSERTED IN THE BUG DATABASE.

http://nagoya.apache.org/bugzilla/show_bug.cgi?id=22513

Scope "leaks" using multiple sessions using new flow framework

           Summary: Scope "leaks" using multiple sessions using new flow
                    framework
           Product: Cocoon 2
           Version: 2.1
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: Major
          Priority: Other
         Component: core
        AssignedTo: dev@cocoon.apache.org
        ReportedBy: g.brand@interface-business.de


Using the new flow framework I encountered numerous null pointer exceptions 
using the cocoon object's getComponent. This only 

happened if more than _one_ session accessed the application.

After adding a lot of debug messages to the FOM_* classes I noticed that the 
FOM_Cocoon instance changed in the flow script. 

After 2 days of tracing I finally found what causes the problem:

I am using javascript objects in my application and somehow the scope changes 
to the scope created by the latest session. This 

scope then is used in _all_ sessions. Obviously all but the last sessions fail, 
because the FOM_Cocoon object is invalidated 

when it is not used, i.e. all internal references are null.
(When I overloaded the get method in FOM_JavaScriptInterpreter.ThreadScope and 
added a debug message, I could see that the 2nd 

argument (the threadScope instance) suddenly changes. I suspect that the top 
level scope is somehow used for all sessions.)

You can reproduce the problem with the attached simple test script.
It displays the global "application" and "cocoon" object's hashcodes.
One time called from the global function test() and the other time from inside 
the application object.

If you have two or more browser windows open, the first will display something 
like this:

global 
stats(): application.hash = 2829043
         application      = 2829043
         cocoon           = 22342265
application 
Application.stats(): this.hash   = 2829043
                     this.cocoon = 22342265 --> calling stats():
stats(): application.hash = 2829043
         application      = 2829043
         cocoon           = 22342265

-------------------------------------------------------------------
And if you load the same application in the second window:
global 
stats(): application.hash = 9089012
         application      = 9089012
         cocoon           = 19257644
application 
Application.stats(): this.hash   = 9089012
                     this.cocoon = 19257644 --> calling stats():

stats(): application.hash = 9089012
         application      = 9089012
         cocoon           = 19257644

--------------------------------------------------------------------
Once you press continue in the first window you get the unexpected result:
global 
stats(): application.hash = 2829043
         application      = 2829043
         cocoon           = 22342265
application 
Application.stats(): this.hash   = 2829043
                     this.cocoon = 22342265 --> calling stats():
stats(): application.hash = 9089012
         application      = 9089012
         cocoon           = 19257644

As you can see the scope changes and stays this way for every called function 
or method. It doesn't matter if I call stats() 

from Application.stats() or output the values directly, both report the wrong 
ones (inside the Application object this!=

application and cocoon is invalid).

Obviously this error does not occur until one is using javascript objects in 
the flow scripts.
I don't know if that is the same reason for the npes I had with the old flow 
engine now and then.


I am using Cocoon 2.1 and JDK 1.4.1_03.

Gunnar