[IPython-user] ipython problem on OS X leopard: non-responsive "u" key.

Barry Wark barrywark@gmail....
Wed Oct 8 15:19:55 CDT 2008

On Wed, Oct 8, 2008 at 12:58 PM, Robert Kern <robert.kern@gmail.com> wrote:
> Barry Wark wrote:
>> I know you've gotten a lot of conflicting advice about which python
>> distribution to use. Just for completeness, let me add my 2 cents and
>> take it as you will: unless you have need for python 2.6 (keeping in
>> mind that neither IPython nor NumPy/SciPy officially support 2.6 yet),
>> stick with the system-installed python on Leopard and delete all of
>> the other distributions. Use setuptools to install packages and there
>> won't be any conflicts between system tools and your newly installed
>> packages (you can use distutils as well, but the newly installed
>> packages will be placed _after_ the system-installed packages on the
>> python path). This is the Apple-sanctioned and blessed way to do
>> Python on OS X.
> Unfortunately, "just use setuptools" is not feasible for some packages, like wx.
> And as someone asked on the other thread, what happens when Apple's tools try to
> import numpy? Do they get the version they are expecting, or do they get your
> newer, possibly incompatible version?

Agreed. Packages that cannot use setuptools are a problem. That list
is (fortunately) decreasing, but for those that depend on them (ETS is
an important example), obviously a non-system python is the only
option. So far, I've been able to avoid these situations using
IPython/Twisted/numpy/scipy/matplotlib, though clearly others MMV.

My understanding, from Bill Bumgarner's comments, is that Apple's
system tools import numpy by version via setuptools, so they should be
"immune" to a newly installed version. Are you under a different

>> As a benefit of using the system python, you gain
>> Dtrace support in the python interpreter (without applying a patch and
>> recompiling the python distribution yourself). You can then use
>> Apple's Instruments application (or dtrace at the command line) to
>> profile and trace executing python code. It's very cool.
> I'd be interested in seeing an example of this. AFAICT, Instruments does not use
> the Python utrace() helper and my attempts to use utrace() with raw dtrace have
> met with segfaults.

I've had trouble too, but have had limited success using Instruments
with an Objective-C app that loads python plugins (compiled using
py2app). I haven't tried it with a pure python app or script.

> --
> Robert Kern
> "I have come to believe that the whole world is an enigma, a harmless enigma
>  that is made terrible by our own mad attempt to interpret it as though it had
>  an underlying truth."
>   -- Umberto Eco
> _______________________________________________
> IPython-user mailing list
> IPython-user@scipy.org
> http://lists.ipython.scipy.org/mailman/listinfo/ipython-user

More information about the IPython-user mailing list