FYI, a quick status update tonight from the testers investigating --
may help Andre investigate:

"!threadstate is showing an active thread state on a thread not
associated with where the original CDoc is created. Each thread state
knows about the objects relying on it and so it stays alive. In this
case the threadstate for the CDoc associated with their dialog is gone.
There is a reference counting issue causing this that we can probably
find with iDNA easily. I'm going to run a trace and a TruScan run
tonight and look for a hit. I'll investigate more in the morning after
the run is done."

He's getting iDNA trace and as soon as I have so update from him I'll
let you know.



--- In opmlsupport@yahoogroups.com, "Dave Winer" <dave@...> wrote:
>
> Basically, the problem with IE7 is not at a level that I can fix, so
> I've posted a follow-up report on the frontierkernel list.
>
> http://tinyurl.com/2gd3zx
>
> I guess now we find out if this is really a community process or not.
>
> Help from Microsoft with this problem would be much appreciated, and
> I'm pretty sure that the only person with enough know-how with the
> kernel fo solve this problem, at this time, is Andre Radke.
>
> Dave
>