> 2. Is it really fruitful to discuss  general design issues along with
> (comparatively) small problems like this one - in the sense of
> alternative ways to fix that problem?

Most definitely. The module went into Python without any review
whatsoever. Nobody (but you) has ever looked at the code in detail.
Only now that I look at it I wonder whether the design of the code
is appropriate.

In this specific case, gpolo suggested a reasonable change to the
proposed patch. You opposed this change, pointing out that this
change contradicts the design behind it. As I think the change gpolo
requested is desirable, it *must* be the design that is wrong (as
it restricts us from doing what I think should be done).

So: if you bring up "the design" as a reason for doing things the
way they are done, expect the design to be challenged.

You might argue that with due process, review should have taken
place before the code was integrated. You might be right, but then
the new turtle module wouldn't have been part of Python 2.6.

Wrt. the specific design issue: I believe that attempts to provide
cross-platform GUI in a simple fashion are doomed to fail. Java
AWT is an extraordinary example of it, but many more libraries
exist that essentially prove that cross-platform GUI is a bad idea;
one may argue that Tk itself is also proof of that (although
it is fairly sophisticated, and not at all simple). The fact that
I have not objected to it earlier is simply because I must have
ignored any claims about cross-platform GUIs.