<div dir="ltr">Hi Emmanuelle<br><div class="gmail_extra"><br><div class="gmail_quote">On 9 August 2016 at 14:44, Emmanuelle Gouillart <span dir="ltr"><<a href="mailto:emmanuelle.gouillart@nsup.org" target="_blank">emmanuelle.gouillart@nsup.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">about backporting, what kind of backporting mechanism do you have in<br>
mind? Merging commits by hand, or something more elaborate?<br></blockquote><div><br></div><div>Yes, I think backporting bug-fixes from time to time and making a point release should suffice. The ports may sometimes require some 3-to-2'ing.<br></div><div> <br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
It may be a silly idea, but can't we do it the other way around, that is<br>
have an experimental branch where developers who wish to write<br>
Python3-only code can work, and in which commits to master are merged if<br>
they are Python 3 compatible? It could give us some time to evaluate<br>
whether most developers want to write code that is not compatible with<br>
Python 2.<br></blockquote><div><br></div></div>I don't think that's a silly idea, but I also don't think that addresses two of the points I raised:<br><br><div>- Remove developer overhead associated with keeping both Python 2 and 3-isms in mind<br><br></div><div>I'd like developers to start thinking in Python 3, not the other way around. Barriers for Python 3 implementations should be removed, not raised, given that that's the direction we'll have to move eventually.<br></div><div><br></div><div></div>- Utilize new language features such as the @-operator and required keyword arguments<br><br></div><div class="gmail_extra">These are not available unless we make Python 3 the default.<br><br></div><div class="gmail_extra">As for whether most developers *want* to write code that is incompatible with Python 2, my guess is that they won't pay much attention. I've moved to Python 3 more than a year ago already, and quite a few of my personal projects already do not run under Python 2---not because I actively pushed for Python 3, just because I didn't explicitly go back and fix it for the 2.7 case. I also can't think of many reasons to justify that effort.<br><br></div><div class="gmail_extra">The only people who should be affected by the suggested change are those stuck on Python 2 but need the latest scikit-image features. But perhaps they can also contribute some of the (fairly lightweight) backporting efforts to help out?<br><br></div><div class="gmail_extra">Best regards<br></div><div class="gmail_extra">Stéfan<br></div></div>