You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@openwebbeans.apache.org by "Eric Covener (JIRA)" <ji...@apache.org> on 2010/06/23 21:14:53 UTC

[jira] Created: (OWB-401) ELContextStore not cleaned up for some JSP EL lookups

ELContextStore not cleaned up for some JSP EL lookups
-----------------------------------------------------

                 Key: OWB-401
                 URL: https://issues.apache.org/jira/browse/OWB-401
             Project: OpenWebBeans
          Issue Type: Bug
          Components: Injection and Lookup
    Affects Versions: M4
         Environment: jetty + guess sample + simple dependent bean + jsp
            Reporter: Eric Covener
            Assignee: Gurkan Erdogdu


Playing with the guess sample with an added dependent bean and a JSP with a simple EL expression referencing the same bean a few times results in 1 dependent bean instance being used for two expressions

openwebbeans.properties:
 org.apache.webbeans.spi.adaptor.ELAdaptor=org.apache.webbeans.el10.EL10Adaptor
+org.apache.webbeans.application.jsp=true

added JSP:

 <c:out value="${dependentBean}"/>
 <c:out value="${dependentBean}"/>

This results in only one bean instance being created, because the first bean is left in the thread-local ELContextStore for the 2nd expression.

Our current code counts on the ELResolver lookup being driven by our own ELValueExpression, which gives us a chance to cleanup the thread-local context on the way out, but this path goes right from jetty to our ELResolverCode twice with no ValueExpression.

This seems to always work correctly from JSF, whether deferred evaluation is used for the value or not.  But this was only tested with the output mechanism in login.xhtml so it may just be good/bad luck.


-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Resolved: (OWB-401) ELContextStore not cleaned up for some JSP EL lookups

Posted by "Mark Struberg (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/OWB-401?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Mark Struberg resolved OWB-401.
-------------------------------

         Assignee: Mark Struberg  (was: Gurkan Erdogdu)
    Fix Version/s: 1.0.0-alpha-2
       Resolution: Fixed

> ELContextStore not cleaned up for some JSP EL lookups
> -----------------------------------------------------
>
>                 Key: OWB-401
>                 URL: https://issues.apache.org/jira/browse/OWB-401
>             Project: OpenWebBeans
>          Issue Type: Bug
>          Components: Injection and Lookup
>    Affects Versions: M4
>         Environment: jetty + guess sample + simple dependent bean + jsp
>            Reporter: Eric Covener
>            Assignee: Mark Struberg
>             Fix For: 1.0.0-alpha-2
>
>   Original Estimate: 4h
>  Remaining Estimate: 4h
>
> Playing with the guess sample with an added dependent bean and a JSP with a simple EL expression referencing the same bean a few times results in 1 dependent bean instance being used for two expressions
> openwebbeans.properties:
>  org.apache.webbeans.spi.adaptor.ELAdaptor=org.apache.webbeans.el10.EL10Adaptor
> +org.apache.webbeans.application.jsp=true
> added JSP:
>  <c:out value="${dependentBean}"/>
>  <c:out value="${dependentBean}"/>
> This results in only one bean instance being created, because the first bean is left in the thread-local ELContextStore for the 2nd expression.
> Our current code counts on the ELResolver lookup being driven by our own ELValueExpression, which gives us a chance to cleanup the thread-local context on the way out, but this path goes right from jetty to our ELResolverCode twice with no ValueExpression.
> This seems to always work correctly from JSF, whether deferred evaluation is used for the value or not.  But this was only tested with the output mechanism in login.xhtml so it may just be good/bad luck.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (OWB-401) ELContextStore not cleaned up for some JSP EL lookups

Posted by "Eric Covener (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/OWB-401?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12881979#action_12881979 ] 

Eric Covener commented on OWB-401:
----------------------------------

As I now understand it, the piece were missing is finding container-specific/non-portable methods to wrap the JSP ExpressionFactory like we do portably for JSF.

> ELContextStore not cleaned up for some JSP EL lookups
> -----------------------------------------------------
>
>                 Key: OWB-401
>                 URL: https://issues.apache.org/jira/browse/OWB-401
>             Project: OpenWebBeans
>          Issue Type: Bug
>          Components: Injection and Lookup
>    Affects Versions: M4
>         Environment: jetty + guess sample + simple dependent bean + jsp
>            Reporter: Eric Covener
>            Assignee: Gurkan Erdogdu
>   Original Estimate: 4h
>  Remaining Estimate: 4h
>
> Playing with the guess sample with an added dependent bean and a JSP with a simple EL expression referencing the same bean a few times results in 1 dependent bean instance being used for two expressions
> openwebbeans.properties:
>  org.apache.webbeans.spi.adaptor.ELAdaptor=org.apache.webbeans.el10.EL10Adaptor
> +org.apache.webbeans.application.jsp=true
> added JSP:
>  <c:out value="${dependentBean}"/>
>  <c:out value="${dependentBean}"/>
> This results in only one bean instance being created, because the first bean is left in the thread-local ELContextStore for the 2nd expression.
> Our current code counts on the ELResolver lookup being driven by our own ELValueExpression, which gives us a chance to cleanup the thread-local context on the way out, but this path goes right from jetty to our ELResolverCode twice with no ValueExpression.
> This seems to always work correctly from JSF, whether deferred evaluation is used for the value or not.  But this was only tested with the output mechanism in login.xhtml so it may just be good/bad luck.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.