Comment 86 for bug 728798

I've found the *real* cause of this bug -- it's a bug in how plugins
are started and stopped.

So there's nothing wrong with the concept of a scheme whitelist for
Java, or (basically) with how my patches for bug 611910 and bug 620773
implemented it. Or if there *is* something wrong with the concept of
a Java scheme whitelist (or with my implementation of it), this bug
doesn't demonstrate it.

I haven't yet worked out the details of the
plugin-starting-and-stopping bug, or found out when it started
happening (or if it still happens on the trunk). That will be at
least several days' work -- which I'll start next week. I'll open a
new bug on it.

I don't normally announce that I've found a bug before working out the
details. But some of us (including myself) have been anxious about
the possibility that using a Java scheme whitelist will cause trouble
on the trunk (which still has my patch for bug 611910). So I just
wanted to relieve that anxiety.

For the curious: Break on NS_CheckIsJavaCompatibleURLString() and try
to reproduce this bug. You'll find that it only gets called once
(when the applet is first loaded, successfully), and that it never
indicates that any urlString is incompatible.