[
https://issues.apache.org/struts/browse/WW-2543?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=43497#action_43497
]
Sami Dalouche commented on WW-2543:
-----------------------------------
I doubt it has anything to do since I have been experiencing the problem with
many versions of Struts (cannot remember when I started using codebehind and
experiencing these problems but I think it dates back to 2.0.9 or 2.0.10) and
Spring, but, I use (most of the details are probably irelevant, but...) :
- Spring 2.5.1
- Struts 2.1.0 / Struts 2.1.1-SNAPSHOT (same problem)
- Hibernate 3.2.5 (which starts all its stuff at startup time, might have an
effect on Struts 2 startup..
- AspectJ 1.5.4 with Spring AspectJ integration
- tons of small libraries and commons-*
- Acegi Security 1.0.5
- Sitemesh 2.3
- PostgreSQL JDBC drivers + PostGIS JDBC extensions
- ehcache 1.3.0
- dwr 2.0.1
- freemarker 2.3.11
I know I use tons of stuff and that would probably help to use less stuff in
order to isolate the problem.. But reproducing the bug is so random that it's
pretty hard to remove/add a feature and tell with confidence that it works /
doesn't work.... So, if other people get the same behavior, please post a
comment with your configuration, versions, and whatever you can add to this....
> Weird Codebehind startup bug (race condition, no magic way to reproduce) :
> switches interceptor stacks
> ------------------------------------------------------------------------------------------------------
>
> Key: WW-2543
> URL: https://issues.apache.org/struts/browse/WW-2543
> Project: Struts 2
> Issue Type: Bug
> Components: Plugin - CodeBehind
> Affects Versions: 2.1.1
> Reporter: Sami Dalouche
>
> Hi,
> I know this bug report is really vague and hard to reproduce, but I thought
> it'd be useful to publicly report the problem so that other people who get
> the same behavior can try to give more details...
> My struts config :
> Several packages defined : "default", "longProcess", "simple", and each
> package has its own default interceptor ref (defaultStack for "simple",
> longProcessStack for "longProcess", and a custom interceptor stack for
> "default".
> If I activate the codebehind, and map different actions to different parent
> packages :
> @ParentPackage("longProcess")
> @ParentPackage("simple")
> @ParentPackage("default")
> And what seems to happen is that randomly, the webapp is sometimes started
> correctly, and sometimes started in a "screwed-up" mode. What "screwed-up"
> mode means is that the actions have a randomly chosen package ("longProcess",
> "simple", or "default". (the visible behaviour is that the wrong interceptor
> stacks are applied )
> sometimes means at least one startup out of ten (mvn clean compile
> tomcat:run, or mvn clean compile jetty:run). Sometimes less, sometimes
> more.... I have the feeling that if I hit the web application soon enough
> right after everything is initialized, I have more chances to boot into
> "screwed-up mode".
> 1] I could not reproduce the problem with codebehind disabled
> 2] I could not reproduce the problem when only one package name is used in
> @ParentPackage
> 3] I have the feeling that the more codebehind-annotated actions there are on
> the webapp, the more chances I have to hit the bug.
> I know this is weird... After relaunching my webapp for a few hundred times,
> I cannot find more information than that....
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
