[Pythonmac-SIG] Building Python with X11 Tkinter

[C. E. Ball]

Hi,

Thanks for helping me out.


On Dec 10, 2007 9:02 AM, Edward Hartley <ed_hartley at mac.com> wrote:
[...]
> PS just seen your latest post IIRC and passing flags is the alternative to
> what I described above and should work unless something in the script is
> detecting the platform and overriding the options you're passing.
> Someone on this list would be able to answer that one I'm sure.

Thanks, I also think something in the script is detecting the platform
and overriding the options...



On Dec 10, 2007 2:56 PM, Ronald Oussoren <ronaldoussoren at mac.com> wrote:
[...]
> There is one tricky bit though: Python's build process "knows" that
> Tcl/Tk on OSX is a framework and won't even look for a unixy library
> location.  That's easy to work around though: edit setup.py and look
> for the definition of detect_tkinter. At the start of that method it
> checks if the platform is 'darwin' and does something osx-specific
> when it is. Just disable that test to get the standard unix behaviour.
>
> That is, replace "        if (platform == 'darwin' and" by "        if
> 0 and (platform == 'darwin' and".

Actually, I tried that early on (I commented out that 'if' statement),
but Python still seemed to end up using the framework Tk. The output
of the build appeared to indicate that the correct Tkinter files had
been found, but when I started Tkinter from Python, it looked like the
Aqua version (comparing it with the X11 version I built by renaming
the Tcl and Tk framework directories).

I guess I have two questions:
(1) How sure are you that making the change above to setup.py is
enough? If you're fairly sure it should be enough, I'll check again
with Python 2.5 from the beginning (I was using Python 2.4).
(2) How can I tell for certain which Tkinter is actually being used by
Python? I feel a bit stupid, but I can't figure out how to get a
definitive path to indicate the location! All I can do is see the
difference between X11 Tkinter and the normal Aqua one.


> BTW. Could you please file a bug report for this at bugs.python.org?

Yes, I'm happy to do that; such an argument would definitely be useful to me.


Chris

(www.topographica.org: an open-source software package for
computational modeling of neural maps)