> Do you mind to share with us what is the cause of the problem?  If it is too
> long to describe, then don't waste your time.

It's quite tricky but ok, I'll try it. It's actually a *great* example why
Smalltalk is so much better than C/C++.

As you may know, I'm re-opening the standard output and error for
redirecting everything from there into a file so that it can be shown upon
Squeak exiting. For this, I created a temporary file and reopened the
standard and error output. Now, the bug was that got the order wrong
(*don't* ask me why ;-))  so that I re-opened the file before trying to
get temporary file name.  This results in actually *freeing* of the handle
since the file could not be created. So far, the standard handle was just
not there (btw, I only did this wrong with the stderr handle so all stuff
going to stdout went ok).

Now comes the funny stuff: *Because* I had actually closed the standard
handle, the system (e.g., the run-time library) felt free to re-use the
resources associated with it. In other words, when a file was opened there
was a certain chance that it would be stored in the place normally
reserved for the standard handle. This *can* lead to all sorts of
unforeseen effects but since most files are closed upon exiting the image
there is usually no problem with it (and that's exactly why you could exit
Squeak by using the desktop menu).

However, the order of operations when doing *particular* this update led
to the effect that during fileIn of the update the changes file was
stored in the place of the standard handle when you tried to exit through 
the close button. And then the cleanup goes wrong since it does not expect
to have a *huge* error output.

So again, thanks for finding the "right" order of operation ;-)

That's it.
  Andreas
-- 
Linear algebra is your friend - Trigonometry is your enemy.
+===== Andreas Raab ============= (raab at isg.cs.uni-magdeburg.de) =====+
I Department of Simulation and Graphics      Phone: +49 391 671 8065  I
I University of Magdeburg, Germany           Fax:   +49 391 671 1164  I
+=============< http://isgwww.cs.uni-magdeburg.de/~raab >=============+