On Tue, May 4, 2010 at 02:14, Roman Kennke <roman.ken...@sun.com> wrote: > I also have a testcase that triggers the problem pretty reliably (at > least on my system). Problem is, it can't be 100% reliable, and it > likely depends on system setup (i.e. fonts installed). Find the testcase > attached.
The important thing is that not that it reliably reproduces the deadlock, but that it occasionally reproduces the deadlock, and that the test itself is portable, that is, it reliably passes with the fix on all the various platforms, including Windows systems and Linux systems that don't have any interesting fonts installed. Anyways, I will go ahead and commit this to openjdk6. A test can always be committed later. Martin > /Roman
