[ITK] SimpleITK Python wheels?

[Bradley Lowekamp]

Hello,

Thank you for your interest with SimpleITK. 

You can read how to install with pip here:
https://itk.org/Wiki/SimpleITK/GettingStarted#Generic_Distribution

For some reason I believe audit wheel changed the version from 0.10.0 to 0.10.0.0:

https://pypi.python.org/pypi/SimpleITK/0.10.0.0

I need to look into the further. 

The problem with the OS X wheels is that they are too big for pipy at about 75mb. These are still fat binaries of i386 and x86_64. I have not check what pip would do with just a x86_64 wheel but with a fat Python version. 

HTH,
Brad

> On Jul 30, 2016, at 6:08 PM, Matt McCormick <matt.mccormick at kitware.com> wrote:
> 
>> On Sat, Jul 30, 2016 at 5:45 PM, Matthew Brett <matthew.brett at gmail.com> wrote:
>> Hi,
>> 
>> On Sat, Jul 30, 2016 at 10:35 PM, Matt McCormick
>> <matt.mccormick at kitware.com> wrote:
>>> Hi Matthew,
>>> 
>>> The just-released SimpleITK 0.10.0 has OSX and manylinux wheels in
>>> addition to Windows wheels.
>> 
>> Ah - yes - I saw 0.10.0 - but the only wheels on pypi are Windows
>> wheels - unless I'm looking at the wrong page:
>> 
>> https://pypi.python.org/pypi/SimpleITK/0.10.0
>> 
>> I couldn't see any 0.10.0 wheels at
>> https://itk.org/SimpleITKDoxygen/html/PyDownloadPage.html or
>> http://www.simpleitk.org//SimpleITK/resources/software.html .
> 
> The wheels are currently on Sourceforge -- we will work on getting
> them on PyPI and updating the documentation:
> 
>  https://sourceforge.net/projects/simpleitk/files/SimpleITK/0.10.0/Python/
> 
> 
>> I did try installing the 0.9.1 wheel via the sourceforge address, but
>> that failed at the download step, so I had to download manually.
>> 
>> I would really help if the wheels were up on pypi - it's generally
>> very reliable and fast, and it's where people are expecting to find
>> them.
>> 
>>> We are working on improved wheel support in SimpleITK and also
>>> hopefully ITK, VTK, symengine, pyne, matplotlib... with the
>>> scikit-build project:
>>> 
>>>  https://github.com/scikit-build/scikit-build
>>> 
>>> We made many strides since the 2016 SciPy conference. Additional
>>> community collaboration and contributions are welcome :-).
>> 
>> Good to know - thanks for the pointer.
>> 
>> Here is my current machinery for building and testing manylinux and
>> OSX wheels : https://github.com/matthew-brett/multibuild - it's
>> currently in use for numpy, scipy, matplotlib, pandas, h5py and
>> others,
> 
> Very cool! Hopefully we can re-use and build upon it.
> 
> Thanks,
> Matt