cxf-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "onder sezgin (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CXF-7011) same method signature needs to be added in Continuation
Date Sat, 20 Aug 2016 09:56:22 GMT

    [ https://issues.apache.org/jira/browse/CXF-7011?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15429313#comment-15429313
] 

onder sezgin commented on CXF-7011:
-----------------------------------

well my intention was not to replace one method ref with another in this highly sensitive
code :)
in resume method the implementatio was like;

{code}
public void resume() {
            isResumed = true;
            isPending = false;
            redispatch();
}
{code}

and in onTimeout callback
resume method was called.

So what i did was to replace the onTimeout callback implementation
as the following

{code}
public void onTimeout(AsyncEvent event) throws IOException {
            isPending = false;
            redispatch();
}
{code}

instead of
{code}
public void onTimeout(AsyncEvent event) throws IOException {
            resume();
}
{code}


so as to avoid 
{code}
isResumed=true
{code}

The reason i believe this is right is that because we had similar execution while using jetty.

After your latest changes, the fix i did in camel-cxf (CAMEL-10171) falls out (for jetty).
(which is against your initial suggestion that we should fix this camel) However, as you set
isResumed=true in 3.1.8 SNAPSHOT, memory leak does not come up again. However, we can not
detect Continuation timeout in camel-cxf and can not mark camel exchange with exchangetimedoutexeption
(which is again against your suggestion that we should fix this camel-cxf, as your latest
change makes memory leak in CAMEL-10171 vanish.) and camel-cxf acts like  as if continuation
resumed, instead camel would be able to detect continuation timeout.

So at this point, your latest change and what i am trying to do gets clashed and does not
completely resolve CAMEL-10171 issue. (before your changes in 3.1.8 SNAPSHOT, what camel-cxf
issue 10171 attempts to do is well right. After your changes, memory leak does not come up
but what we intent in camel-cxf falls out)  

i updated my PR, https://github.com/apache/cxf/pull/159 (assuming you have not done the change
for 3.1.8)

However, apparently, your latest change requires another fix in camel-cxf 

[~davsclaus] and [~ddms], could you please bring your thoughts on what we have in camel-cxf
component?

And lets have it right at the both ends(cxf-rt-transport-http   /    cxf-rt-transport-http-jetty
and camel-cxf)

Thanks

> same method signature needs to be added in Continuation
> -------------------------------------------------------
>
>                 Key: CXF-7011
>                 URL: https://issues.apache.org/jira/browse/CXF-7011
>             Project: CXF
>          Issue Type: Sub-task
>          Components: Transports
>    Affects Versions: 3.1.7
>            Reporter: onder sezgin
>             Fix For: 3.2.0
>
>
> As this method requires access over Continuation interface this should be provided and
other implementations like transport-jms and transport-http with  default implementation and
transport-http-jetty with real implemenration should be provided.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message