Unable to run IDLE from 2.3?

Kendall Clark kendall at monkeyfist.com
Tue Aug 5 21:30:04 CEST 2003

On Tue, Aug 05, 2003 at 03:14:22PM -0400, Tim Peters wrote:
> In addition to the two possible causes Alex spelled out, another common
> problem is some *other* program setting up Tcl or Tk environment variables
> pointing to an inappropriate version of Tcl/Tk.  The installer for Ruby is
> the most common culprit for this problem, so if you've installed Ruby that's
> probably the cause.

I'm battling what seems to be a related issue on RH 9 (accursed Red Hat!). I
installed 2.3 from src tarball last week, and it worked fine. The new Idle
is very sexy. Yay, idle.

Then last night I decided to upgrade from RH 8 to RH 9 (accursed Red Hat!).
That was a disaster on many fronts (it's not a good thing to have corrup CD
media on the 3rd disc of a KRUD update...argh!), but it also totally trashed
my 2.3's _tkinter.so. Now when I try to rebuild 2.3 from src, I can't seem
to get a _tkinter.so which points to valid tcl & tk libraries (despite using
"--with-libs="" to point to the libtcl & libtk RH 9 installed).

(The extent of my dynamic library fu on Linux is to futz around with
symlinks from the places where _tkinter.so wants to find tcl/tk libs to the
places where they actually are on my RH 9 (accursed Red Hat!) system, then
to rerun /sbin/ldconfig. This "trick" didn't work either.)

So I decided that RH is completely evil (accursed Red Hat!), and I would
just install Tk & Tcl from scratch in /usr/local/lib and rebuild 2.3 using
those libs, like this:

     ./configure  --with-libs="/usr/local/lib/libtcl8.4.so \

This works, but goddam what a pain in my ass...In case it's now clear by
now: I hate Red Hat (though hitting back at SCO is pretty cool, trying to
keep up with Python on a RH box has been way too painful for far too long).

I recommend repeated playings, at max volume, of Bob Dylan's "Idiot Wind" if
you have to fight this battle -- it seemed to help my mood.

Kendall Clark
Jazz is only what you are. -- Louis Armstrong

More information about the Python-list mailing list