[Python-ideas] Re: [Distutils] Re: Linux binary wheels?

2019-08-20 Thread Nick Timkovich
On Tue, Aug 20, 2019, at 5:05 AM Matthew Brett 
wrote:

> ...  Unless you meant wheels for non-Intel platforms, in which case,
> please do say more about you need.


Minor tangent: I've seen some people use https://www.piwheels.org/ for
Raspberry Pi (ARM 6/7), but could the ARM binaries be uploaded to PyPI?

I think I'm conflating the wheel building spec (is manylinux amd64
specific, or as long as the libraries are on any architecture?),
toolchains, environment (sounds like Piwheels provides a platform to build
them on), and package hosting (can PyPI host arbitrary archs?) in that one
sentence.
___
Python-ideas mailing list -- python-ideas@python.org
To unsubscribe send an email to python-ideas-le...@python.org
https://mail.python.org/mailman3/lists/python-ideas.python.org/
Message archived at 
https://mail.python.org/archives/list/python-ideas@python.org/message/OXSUW73EO5DTUO34EFURN3KHCDAKNS4Z/
Code of Conduct: http://python.org/psf/codeofconduct/


[Python-ideas] Re: [Distutils] Re: Linux binary wheels?

2019-08-20 Thread Paul Moore
On Tue, 20 Aug 2019 at 14:50, Brian Skinn  wrote:

> I wonder if there's an OS dependence here, though -- I'm almost certain I've 
> had to use `--only-binary` in the past, to avoid pip on my Windows machines 
> trying to download and build sdists, even when wheels were available.

Pip prefers newer versions over older ones. If there's a newer version
with no binaries, pip will use the source for that version and try to
build, even if there are older binaries. `--only-binary` would address
that (although there is also a `--prefer-binary` flag in newer
versions of pip, which is better suited to that situation).

Paul
___
Python-ideas mailing list -- python-ideas@python.org
To unsubscribe send an email to python-ideas-le...@python.org
https://mail.python.org/mailman3/lists/python-ideas.python.org/
Message archived at 
https://mail.python.org/archives/list/python-ideas@python.org/message/FR464FPXMZKTR5QBCWYTYSPBSOXFRKYM/
Code of Conduct: http://python.org/psf/codeofconduct/


[Python-ideas] Re: [Distutils] Re: Linux binary wheels?

2019-08-20 Thread Brian Skinn
On Tue, Aug 20, 2019 at 6:14 AM Matthew Brett 
wrote:

> Hi,
>
> On Tue, Aug 20, 2019 at 11:10 AM Brian Skinn 
> wrote:
> >
> >
> >
> > On Tue, Aug 20, 2019, 06:05 Matthew Brett 
> wrote:
> >>
> >> Hi,
> >
> >
> >> See the links that Wes posted for more details, or try pip installing
> >> Numpy and Scipy in a new virtualenv, and see what happens.
> >
> >
> > Probably needs to:
> >
> >pip install --only-binary :all: numpy scipy
>
> Pip doesn't build from source by default - here's the output from my
> machine just now:
>
>  $ pip install numpy scipy
> Collecting numpy
>   Downloading
> https://files.pythonhosted.org/packages/4a/2e/cf0a2fea6d97604a0e058e804b50d589c31b360b317be9f5c126b22a560e/numpy-1.17.0-cp35-cp35m-macosx_10_6_intel.macosx_10_9_intel.macosx_10_9_x86_64.macosx_10_10_intel.macosx_10_10_x86_64.whl
> (14.9MB)
>  || 15.0MB 38.9MB/s
> Collecting scipy
>   Downloading
> https://files.pythonhosted.org/packages/c5/40/f73bc951b060ba143f2092a3a2dd755f5ef2b52376fc3f4d297942aef04b/scipy-1.3.1-cp35-cp35m-macosx_10_6_intel.macosx_10_9_intel.macosx_10_9_x86_64.macosx_10_10_intel.macosx_10_10_x86_64.whl
> (27.6MB)
>  || 27.6MB 35.8MB/s
> Installing collected packages: numpy, scipy
> Successfully installed numpy-1.17.0 scipy-1.3.1
>
>
Hard to argue with concrete data -- my bad!

I wonder if there's an OS dependence here, though -- I'm almost certain
I've had to use `--only-binary` in the past, to avoid pip on my Windows
machines trying to download and build sdists, even when wheels were
available.



> Cheers,
>
> Mattthew
>
___
Python-ideas mailing list -- python-ideas@python.org
To unsubscribe send an email to python-ideas-le...@python.org
https://mail.python.org/mailman3/lists/python-ideas.python.org/
Message archived at 
https://mail.python.org/archives/list/python-ideas@python.org/message/OD7TDQFPTOZOST6FP367KG5LEWQJ5VES/
Code of Conduct: http://python.org/psf/codeofconduct/