Idle loops - sometimes a little too idle?

Hi Blair,

There's every possibility that the problem I'm having is due either to my
own fiddlings or the fact that I'm working with a relatively ancient
installation of NT that's customized (by our hospital computer guys) to the
extreme.  With that said, I have a situation that appears to be case of
"move the mouse and things happen; don't move it and things don't happen".

I noticed something along these lines, though I thought less severe, on the
first NT machine that I deployed.  IIRC, I do some mouse jiggling to get
through startup faster (rather than at all) and then things run fairly
normally.  The latter part is true - it does run just fine, and there's lots
of background processing.  It also was set up by our department rather than
the hospital, and I suspect more recently, though I have yet to compare
versions.

You might recall that I've found it helpful to shorten the timeouts in the
idle loop on 9x.  We concluded some time ago that the system was going idle
in good faith, shortly after which events that did not result in awakening
were arriving.  For NT, I did something similar.  What I have yet to verify
is that the machine that's working is running such a modified idle loop.  If
it is, or if it tolerates one that I hope to test on it tomorrow, then I'd
be left asking what I should check in terms of service packs, versions, etc.
to get the other machine moving.

I hope there's a good question in there somewhere :)  If any of this sounds
familiar, please feel free to suggest the obvious: I'm pretty green with NT.

Have a good one,

Bill

--
Wilhelm K. Schwab, Ph.D.
[hidden email]

Hi Blair,

> I noticed something along these lines, though I thought less severe, on
the
> first NT machine that I deployed.  IIRC, I do some mouse jiggling to get
> through startup faster (rather than at all) and then things run fairly
> normally.  The latter part is true - it does run just fine, and there's
lots
> of background processing.  It also was set up by our department rather
than
> the hospital, and I suspect more recently, though I have yet to compare
> versions.

The situation is a little worse/different than I remembered it.  If I work
really hard to keep the mouse still during app startup, the "working" app
will in fact just sit there refusing to do anything; after that, it's ok.  I
missed the startup problem because of the chaotic envinroment and the
resulting non-ergonomic workstation.

I first noticed similar problems a couple of years ago on Win9x, and have
tweaked #idle95 ever since.  The original problem involved COM client/server
pairs that would stall out for intervals that turned out to be controllable
by the timeout in the #idle95.  The current "working" app appears to stall
(on NT) on a socket read; by that, I mean that only thing to do is to wake
up to read from a socket, but, the app never wakes up, or at least not
enough to notice or read the socket data.

Not having much to lose, I made a backup and then changed #idleNT to do the
same thing as #idle95.  The resulting deployed app ran as expected on the
target NT box.

Have a good one,

Bill

--
Wilhelm K. Schwab, Ph.D.
[hidden email]