Dropping support for 2.7 in 2020

Emmanuelle Gouillart emmanuelle.gouillart at nsup.org
Wed May 18 02:47:19 EDT 2016


At the moment, can you estimate the cost of supporting Python 2.7, in
terms of additional code? (I would say that testing 2.7 in CI is not a
true cost, since it represents server time and not human time) 

Best,
Emma

On Wed, May 18, 2016 at 06:32:14AM +0000, Stéfan van der Walt wrote:
> Hi Johannes

> On Tue, 17 May 2016 at 23:26 Johannes Schönberger <jsch at demuc.de> wrote:

>     Even though official support for 2.7 will be dropped on the Python side, my
>     feeling is that it will stick around in official linux repositories for
>     quite some time longer. Especially on clusters etc., as you mentioned. I
>     would suggest to drop 2.7 support as soon as Numpy/Scipy drops it - this is
>     an upper bound as we depend on it and, at the same time, numpy/scipy are
>     important enough libraries that should also convince the last developer to
>     make the switch to 3.x.


> It is true that Python 2.7 will stick around beyond 2020, but it will not
> receive bug-fixes any longer.  scikit-image 0.18 (or whatever :) will also
> stick around along with 2.7, and we can keep supporting that release.  But for
> any new releases beyond that point it makes little sense to support 2.7.

> Stéfan



More information about the scikit-image mailing list