On Thu, Apr 9, 2009 at 5:04 PM, Serge Baranov <[email protected]> wrote:
> Hello Emmanuel,
>
> E> Or maybe the foobar attribute is not anymore present in the session.
>
> That's was my suggestion in the very first mail. It can be null only
> if the order guaranteed by the OrderedThreadPoolExecutor is broken
> now.

Can you post the chain you are using ?

> E> What if you do something like :
> E> Object foobar = session.getAttribute( foobar ) ?
>
> I can't experiment right now as the server needs to be running. All I
> can tell is that it was working fine before, started throwing
> exceptions after MINA update, and now is working again after rolling
> back to the older MINA version.
>
> There may be something wrong with the latest commits and the code
> needs a review.

This is a possibility. Saddly enough, the last commit also fix a
really bad issue (the usage of a non thread-safe data structure for
the write chain).

I will review the commit.

It's also possible that the commit didn't broke MINA, but has side
effect and expose another bug elswhere in the code...

Not simple :/

>
> --
> Best regards,
>  Serge
>
>



-- 
Regards,
Cordialement,
Emmanuel Lécharny
www.iktek.com

Reply via email to