[Bf-committers] numpy for 2.70

[dbaker@citytech.cuny.edu]

Would it be useful for me to get the binaries out of my anaconda installs
and send them? I have OS X, Ubuntu and Windows 7 machines in my lab and
students who need practice working with anaconda that I can put on the task
today. I can make certain they don't use the mkl versions.
On Feb 24, 2014 3:25 PM, "dbaker at citytech.cuny.edu" <
damonlorenbaker at gmail.com> wrote:

> Anaconda supports python 3.3,  it just defaults to 2.7 To make a 3.3
> environment you have to use their package management tool, conda. (you can
> skip having to download the monster sized installer of 2.7 by using
> miniconda which is just a script that installs a minimal conda system and
> lets you decide which pieces you want, including setting up a 3.3 only
> environment) here is a list of all the packages it supports for each
> version of python http://docs.continuum.io/anaconda/pkgs.html numpy 1.8.0
> is in there.
>
>
> On Mon, Feb 24, 2014 at 1:36 PM, CoDEmanX <codemanx at gmx.de> wrote:
>
>> They charge for MKL, so the free version should definately not use /
>> require it.
>>
>> Direct download links for Anaconda Windows builds:
>>
>> http://repo.continuum.io/archive/.winzip/
>>
>> I checked the content in 7zip and it seems to contain py2.7 builds only?
>> Or is it compatible with py3.3?
>>
>>
>>
>> Am 24.02.2014 18:43, schrieb dbaker at citytech.cuny.edu:
>> > could we use one of the builds of numpy from anaconda
>> > https://store.continuum.io/cshop/anaconda/ I believe their free
>> version (
>> > the one you get by default if you install it using conda)  isn't built
>> > against MKL.
>> >
>> >
>> > On Mon, Feb 24, 2014 at 9:48 AM, Martijn Berger <
>> martijn.berger at gmail.com>wrote:
>> >
>> >> I have tried compiling numpy but it seems you need either gcc/gfortran
>> when
>> >> building for mingw or intel's compiler when building for msvc build
>> python.
>> >> There is a version available online that works at:
>> >> http://www.lfd.uci.edu/~gohlke/pythonlibs/#numpy.
>> >> But there is also a problem. This version is build against the Intel's
>> MKL
>> >> (http://software.intel.com/en-us/intel-mkl) a non free but very high
>> >> performance math library.
>> >> For me the easiest solution would by bundling this version but we it
>> seems
>> >> we can not as that would constitute a GPL violation.
>> >>
>> >> It seems we have 3 options.
>> >>    - we don't bundle numpy on windows
>> >>    - we bundle this numpy and we might violate the GPL
>> >>    - we get a copy of Intel's Fortran compiler so we can build and
>> maintain
>> >> a version of numpy for use with blender that does not violate the GPL.
>> >>
>> >> I am not a layer and not a numpy specialist, any conclusions drawn
>> above
>> >> are the result of my best understanding of our situation
>> >>
>> >>
>> >> On Mon, Feb 24, 2014 at 1:16 AM, Campbell Barton <ideasman42 at gmail.com
>> >>> wrote:
>> >>
>> >>> A while ago it was agreed that blender would bundle numpy, with very
>> >>> positive response from some script authors and general agreement,
>> >>>
>> >>> See meeting minutes:
>> >>>
>> http://lists.blender.org/pipermail/bf-committers/2013-April/039809.html
>> >>>
>> >>> Previous discussions on the topic:
>> >>>
>> >>
>> http://lists.blender.org/pipermail/bf-committers/2012-November/038215.html
>> >>>
>> http://lists.blender.org/pipermail/bf-committers/2012-April/036428.html
>> >>>
>> >>>
>> >>> We setup scons and cmake to bundle numpy with Python, and now the 2.70
>> >>> Linux build bundles numpy 1.80, but (unless I'm mistaken). OSX and
>> >>> ms-windows still don't include numpy.
>> >>>
>> >>> If there are no blocking issues - (we can't get numpy to compile for
>> >> eg...)
>> >>> I think its a reasonable target to have numpy bundled on all platforms
>> >>> for 2.70, not just for Linux.
>> >>>
>> >>> Added tracker tickets for this so this isn't overlooked.
>> >>> https://developer.blender.org/T38791
>> >>> https://developer.blender.org/T38792
>> >>>
>> >>> Maintaining platform deps is often done by volunteers, so if you don't
>> >>> have time for it, say so and someone else can try to fill in the gaps.
>> >>>
>> >>> --
>> >>> - Campbell
>> >>> _______________________________________________
>> >>> Bf-committers mailing list
>> >>> Bf-committers at blender.org
>> >>> http://lists.blender.org/mailman/listinfo/bf-committers
>> >>>
>> >> _______________________________________________
>> >> Bf-committers mailing list
>> >> Bf-committers at blender.org
>> >> http://lists.blender.org/mailman/listinfo/bf-committers
>> >>
>> > _______________________________________________
>> > Bf-committers mailing list
>> > Bf-committers at blender.org
>> > http://lists.blender.org/mailman/listinfo/bf-committers
>> >
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-committers
>>
>
>