At 10:56 PM +0000 02/12/09, Ned Deily wrote:
>Ned Deily <nad@acm.org> added the comment:
>
>FWIW, I am not able to reproduce this using a release3.0 IDLE (so with
>today's patches) built with the default Apple-supplied Tcl/Tk in 10.5. 
>Are you using a newer Tcl/Tk?  Does it happen if you move your current
>preferences out of ~/.idlerc?
>

It does still happen if I move my preferences. And move my .Idle.py.
I did build it with a newer Tk -- 8.5 installed as a Framework.
(And I switched the order of /System/Library/Frameworks and /Library/Frameworks,
which makes more sense. I thought I had seen that as an Issue, but it must have been
just a web page where I saw it. Similarly, I added /opt/local/lib and /opt/local/include
in detect_modules before the directories that are added, so that it finds the readline I
installed with "port".)

So, I started again with a fresh checkout. I didn't switch the order of the Frameworks
directories, and I tried it with and without the directories for readline. I started Python
manually, imported Tk, checked it's version, etc. to confirm I was getting the Apple
installation.  If I try to run bin/idle from the 3.1 Frameworks version directory
I get a segmentation fault. I've seen a lot of weird behavior over the years in my
many configuration and installation escapades across a wide variety of software,
but I've never seen a seg fault with anything to do with Python.  I have the "do you
want to report this" backtrace if you're interested. A little later I'll try this from scratch
on a clean, though PowerPC (G4), machine and see what happens.