We have an application we monitor via an older activex control, from which
we receive event notifications - form opened, user logged in/out, about to
exit, that sort of thing. We poll to detect the application startup using a
'connect' method on the (our application) Com object, which throws an
exception when connect fails - then we loop back round (more on this later)
with a small nap between polls.

We have deployed to machines with an SOE consisting of win2k, + word 2k, and
on these machines we find that, with word running (and our monitoring app
running) and when we start / stop / start our monitored app (and so we are
installing, uninstalling, installing event handlers for the events mentioned
above, as well as handling said events) we find that ms word consumes 100%
cpu.  We are *not* doing any specific word or other office interop calls -
we are talking only to our own com component when we observe this behaviour.

We initially responded by removing our connect polling - we now wait for the
process (our monitored application) to appear in the process list, then
connect, so we are not actively polling through our com object.

That *seemed* to solve the problem for a while. however it now appears in a
new guise - with the series of actions described above( word running, our
app start/stop/start) resulting in our app being non-responsive accross the
com interface - we never return from some com calls.

We have found that an upgrade from word 2000 to word XP solves the problem,
however our clients are reluctant to upgrade the large number of users.

I have not been able to locate a knowlege base article or similar that
describes this - does anyone else have any experience with similar issues?

Thanks in advance,
Brendan.