[icedtea-web] RFC: Backing out patch causing regressions

[Omair Majid]

On 03/05/2012 10:37 AM, Deepak Bhole wrote:
> Hi,
> 
> http://icedtea.classpath.org/hg/icedtea-web/rev/6b46e55a8854
> 
> Is causing regressions and holding back release of 1.2. Omair has taken
> a deep look at the patch and concluded that backing out is the best
> option as fixing requires a significant change.
> 
> I am in agreement -- posting here for the record.
> 
> OK for back-out from 1.2?
> 

Given the regressions caused by this patch, I am okay with backing this
out for now.

I have two comments on this, though:

1. Synchronization of Swing code. This looks rather wrong. If all swing
calls are happening on the EDT, there is nothing to synchronize. If
calls are not happening on the EDT, then there is a deeper problem that
we should solve instead.

2. The change from JFrame.setVisible(false) to JFrame.dispose() is
causing the EDT to shutdown - which is what we want. The trouble is that
when it starts back up, a lot of things can go wrong. There are two
issues I am aware of. One is that the AppContext associated with the EDT
has to be set properly (it's currently not being set, see the thread
starting at [1]). The other is that we will try to query the ThreadGroup
associated with the jnlp applets after they have stop()ed which causes
an IllegalStateException (an assertion, really) in icedtea-web code.

I think we can fix both of these issues properly in the long term.

Thanks,
Omair

[1]
http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2012-March/017500.html