On 19/05/2008, Niklas Gustavsson <[EMAIL PROTECTED]> wrote:
> On Sat, May 17, 2008 at 11:59 PM, Rory Winston <[EMAIL PROTECTED]> wrote:
>  > Are we ready to cut another RC? I can do so if you guys are happy with 
> where
>  > we're at.
>
>
> I still got this issue:
>  https://issues.apache.org/jira/browse/NET-217
>
>  I haven't look into any depth at why it's happening, but changing the
>  test to using unique ports for each of the tests (6900, 6901, 6902...)
>  make JUnit all happy. Should I commit this fix? With it, I would be
>  happy to see 2.0 released.

I think I saw the same error, but only once, so I assumed it was
something else happening on my system.

If it's occurring regularly, then there may be a problem with
releasing the port from previous tests in the same class.

However, I have just seen a problem in the TFTPServer code - the
constructors call launch() which creates a new thread and starts it.
This is not thread-safe, as the class is not guaranteed to be fully
constructed until after the constructor has completed. There are some
other thread-safety problems too; I'll raise a JIRA.

I don't know if it relates to the test failure or not - I suppose the
unsafe constructor could mean that shutdown does not happen reliably.

>
>
>  /niklas
>
>
>  ---------------------------------------------------------------------
>  To unsubscribe, e-mail: [EMAIL PROTECTED]
>  For additional commands, e-mail: [EMAIL PROTECTED]
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to