[Numpy-discussion] 1.8.1 release

Matthew Brett matthew.brett at gmail.com
Wed Feb 26 17:48:31 EST 2014


On Wed, Feb 26, 2014 at 2:33 PM, Chris Barker <chris.barker at noaa.gov> wrote:
> On Wed, Feb 26, 2014 at 11:34 AM, Matthew Brett <matthew.brett at gmail.com>
> wrote:
>> >   - convention is the other option:
>> >     - use binary wheel for in-house deplyment to similar systems
>> >     - use binary wheels for a well-defined python build:
>> >        - for PyPi, that's the python.org builds for Windows and OS-
>> Thanks - that is a very useful summary.
>> It would make sense I think to provide numpy wheels like mine via pypi
>> - as pyzmq does for example.
> Indeed -- and I really appreciate your efforts on this -- I think we should
> be able to get the whole "stack" up there pretty soon (though there is an
> issue with iPython and readline...) Ralf had put together a test set of
> these, too a little while ago.
>> In this case, I believe (Chris correct me if I'm wrong) that someone
>> running via system python would get the usual compile / install, but
>> someone running python.org python would get a near instant numpy,
> That's the idea -- though not entirely sure how that would go without
> testing.

It's currently working for pyzmq - so I supect it would work.

> Also, I think with pip, you need to tell it to look for binary wheels -- it
> won't do that by default.
> pip install --use-wheel numpy

I think --use-wheel is the default for the latest pip ...

>> so that seems like a clear win.
> Agreed. The trick is that it's reasonable for users of Apple's python build
> to want this too -- but I don't know how we can hope to provide that.

We don't support system python for the mpkg, so I think it's
reasonable to leave this little gift for our fellow python.org

In that case, the OSX instructions could (within the next few months)
be as simple as:

Install python from binary installer at python.org
curl -O https://raw.github.com/pypa/pip/master/contrib/get-pip.py
python get-pip.py
pip install scipy-stack

or similar.

> (and macports, and homebrew... but those I feel better about requiring to
> build your own -- really, that's what those systems are designed to do)

Yes, that seems right to me.



More information about the NumPy-Discussion mailing list