Re: [f2py] Can't get Enthought python to work with f2py

Thanks for the clear explanation.

I guess I really do need five fortran compilers! ( and to change my  
path depending if I'm using fink or enthought. )  ugly.



On Jun 30, 2009, at 2:45 PM, Robert Kern wrote:

> On 2009-06-30 14:40, charlie strauss wrote:
>>
>> On Jun 30, 2009, at 12:35 PM, Robert Kern wrote:
>>
>>> On 2009-06-30 12:58, charlie strauss wrote:
>>>> I just installed the latest enthought python distro and now my f2py
>>>> won't compile.
>>>>
>>>> I think the problem is that the enthought f2py is trying to use  
>>>> the fink
>>>> or macports gfortran. But I don't know how to tell it not to.
>>>>
>>>> if I run ipython from fink then the programs do compile.
>>>
>>> You almost certainly shouldn't be running f2py from ipython.
>>
>>
>> uh why not?
>>
>>>
>>>> the specific error that occurs from the enthought f2py is that it  
>>>> does
>>>> not like the --arch command line flag that f2py places on the  
>>>> compile
>>>> command. So another way to try solving this is to somehow adjust  
>>>> the
>>>> default command line args in enthought's f2py so the fink  
>>>> gfortran is
>>>> happy.
>>>
>>> Note that it's not "Enthought's f2py". It's the standard f2py from
>>> numpy. I suspect that Fink is patching numpy to work with its  
>>> standard
>>> gfortran configuration out-of-box.
>>>
>>>> I have also tries using g95 as the -fcompiler=g95 or -- 
>>>> fcompliler=gnu95
>>>> and this also produced different but I think fundamentally simmilar
>>>> errors.
>>>>
>>>> So my first question is how do I get enthought's f2py to stop  
>>>> using the
>>>> fink or macports f2py???
>>>
>>> Make sure that the gfortran that is first on your $PATH is the one  
>>> you
>>> want to use. I recommend the one here:
>>
>> well the enthought python framework is first on the path (as I  
>> showed in
>> the original post). when I invoke ipython or python it is what  
>> launches
>> not the one in fink which is later on the path.
>>
>> So I repeat my original question, how do I tell the enthought  
>> numpy.f2py
>> that it needs to use whatever fortran compiler is used by enthought
>> rather than finks.
>>
>> perhaps the mistake I'm making here is that I'm assuming enthough's
>> distro has a fortran compiler lurking in it for numpy.f2py to use.
>
> Correct. It does not.
>
>> So if I'm going to install a different one, then how do I "patch"  
>> f2py's
>> default arguments so it works with that compiler. Clearly the  
>> gfortran,
>> g9g and gnu95 that come with fink and macports do not work unpatched
>> with enthought's numpy.f2py.
>
> Install the one I recommended. Make sure it is first on the $PATH.
>
> Actually, I was probably incorrect about Fink's patching numpy. They  
> probably didn't. The --arch flags are most likely coming from  
> Python's configuration, as with any extension module. The version of  
> EPD that you installed compiled Python as a Universal binary, so all  
> of the extension modules have --arch flags in them. Fink's Python is  
> probably Intel-only and was not compiled with those flags.
>
>> you are suggesting I install yet another fortran compiler to solve  
>> this
>> problem when I already have at least four installed including two
>> versions of gfortran. This seems kind of a brute force suggestion.
>
> Sorry. Fink's gfortran does not incorporate the Apple patches to  
> work with the --arch flags that are standard on OS X. If one of the  
> four you have installed does incorporate those patches, then go  
> ahead and use that. In my experience, only the one I recommended does.
>
> -- 
> 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
>
>
> _______________________________________________
> f2py-users mailing list
> f2py-users-Y4l6ocDipWCuvFJfX82//[email protected]
> http://cens.ioc.ee/mailman/listinfo/f2py-users

Charlie Strauss
Bioscience Division
[email protected]
505 665 4838
Quidquid latine dictum sit, altum sonatur.

_______________________________________________
f2py-users mailing list
f2py-users-Y4l6ocDipWCuvFJfX82//[email protected]
http://cens.ioc.ee/mailman/listinfo/f2py-users