My 3.2alpha image (4599) doesn't have a Player56 class in it.  This is
probably a new Morphic player which you created (and possibly added to
the flap) which is causing problems.  You might want to browse that
class and look at its superclasses to see if it gives any hints of what
it is.  You could also try deleting the class.  (You may need to "chase
pointers" on the class in an inspector to see if an instance of
something is using it.)

In any case, I always keep a fresh (unmodified) copy of the downloaded
image/changes files around, and save my modified images under different
names.  Then you can always go back to your fresh 4441 image and
update-from-server with that one instead, if you had a problem with your
custom image.

- Doug Way
  dway at riskmetrics.com


Ross Boylan wrote:
> 
> [I tried to post this a couple of days ago, but it looks as if it
> didn't go.]
> 
> I am unable to take in the changes from the servers because at
> 4483makeThumbnailTweak2-di I get an error in the poststscript,
> specifically, doing Flaps replaceWidgetsFlap.  This leads to
> Player56(Model)>>veryDeepFixupWith:, which finds that
> originalDependents is nil and throws an exception.
> 
> My original message had a lovely list of suspects; here is my
> attempted recreation.  There must be something peculiar to my
> environment that is doing this, since it's an old update and there
> were no other complaints I noticed.
> 
> 1. My daughter dragged something weird onto the widgets flap, or did
> something weird to it (she loves the bouncing atoms).  However,
> nothing looks unusual there.
> 
> 2. Something about taking this in with a bunch of earlier updates (I
> started at 4441) causes trouble.
> 
> 3. My VM, really 3.0 on Debian GNU/Linux (though I'm at 3.2 alpha with
> the image) is the culprit.
> 
> 4. The flap I created (which is not a part bin or anything special) is
> really the source of the problem.
> 
> 5. Some local change I've made (tweaks to the publishing process come
> to mind, though they were fixes) interacts badly with the update.
> 
> This assumes the debugger info is accurate, and on some attempts (this
> failure is repeatable) all my instance variables have been nil.  So
> maybe the info is not accurate.