+1 Thanks Jarek for the background information :-)
Cheers, Fokko Op wo 13 nov. 2019 om 01:14 schreef Tao Feng <[email protected]> > +1 > > On Tue, Nov 12, 2019 at 3:31 PM Jarek Potiuk <[email protected]> > wrote: > > > Well. It seems then that my Buster change might also contain dropping 3.5 > > from Travis and switching to 3.6 everywhere we use 3.5 as default version > > :). Good ! > > > > And as soon as we get faster CI we can add 3.8 to the list instead :). > > > > J. > > > > On Wed, Nov 13, 2019 at 12:21 AM Kevin Yang <[email protected]> wrote: > > > > > +1 > > > > > > Ash Berlin-Taylor <[email protected]>于2019年11月12日 周二下午2:51写道： > > > > > > > +1 from me too, especially now Debian ships a newer version. > > > > > > > > And yes, doing it for 2.0 seems like the best time > > > > > > > > -ash > > > > > > > > On 12 November 2019 22:45:42 GMT, Felix Uellendall > > > <[email protected]> > > > > wrote: > > > > >Absolutely +1. In my opinion that should come together with the drop > > of > > > > >Python 2.7. So I mean target that for 2.0 as well. WDYT? Or is it > more > > > > >like sth. for 2.1? > > > > > > > > > >Felix > > > > > > > > > >Sent from ProtonMail Mobile > > > > > > > > > >On Tue, Nov 12, 2019 at 23:02, Kamil Breguła > > > > ><[email protected]> wrote: > > > > > > > > > >> +1 > > > > >> > > > > >> On Tue, Nov 12, 2019 at 10:48 PM Kaxil Naik <[email protected]> > > > > >wrote: > > > > >>> > > > > >>> +1 > > > > >>> > > > > >>> On Tue, Nov 12, 2019 at 9:47 PM Dan Davydov > > > > ><[email protected]> > > > > >>> wrote: > > > > >>> > > > > >>> > +1 > > > > >>> > > > > > >>> > On Tue, Nov 12, 2019 at 4:46 PM Jarek Potiuk > > > > ><[email protected]> > > > > >>> > wrote: > > > > >>> > > > > > >>> > > Yep. It was actually a '+1' in disguise Bolke :). > > > > >>> > > > > > > >>> > > On Tue, Nov 12, 2019 at 10:44 PM Christian Lellmann > > > > >>> > > <[email protected]> wrote: > > > > >>> > > > > > > >>> > > > +1 from my side too! > > > > >>> > > > > > > > >>> > > > Bolke de Bruin <[email protected]> schrieb am Di., 12. > Nov. > > > > >2019, > > > > >>> > 22:39: > > > > >>> > > > > > > > >>> > > > > I guess thats a +1 then? :-) > > > > >>> > > > > > > > > >>> > > > > Op di 12 nov. 2019 22:36 schreef Jarek Potiuk < > > > > >>> > > [email protected] > > > > >>> > > > >: > > > > >>> > > > > > > > > >>> > > > > > I'd love to get rid of python 3.5 > > > > >>> > > > > > > > > > >>> > > > > > We discussed it a few times but as of recently the main > > > > >problem was > > > > >>> > > > that > > > > >>> > > > > > 3.5 was default python version of Stretch debian LTS > (9). > > > > >However > > > > >>> > we > > > > >>> > > > are > > > > >>> > > > > > soon - this week likely - switching to Buster debian > LTS > > > > >as base > > > > >>> > > image > > > > >>> > > > > for > > > > >>> > > > > > our CI tests (and production image that follows) (PR > > > > >>> > > > > > https://issues.apache.org/jira/browse/AIRFLOW-5842). > > > > >Buster comes > > > > >>> > > with > > > > >>> > > > > 3.7 > > > > >>> > > > > > by default and is already 4 months old. > > > > >>> > > > > > > > > > >>> > > > > > I think we should make an educated decision, based on > > what > > > > >we can > > > > >>> > > gain > > > > >>> > > > by > > > > >>> > > > > > moving to 3.6. > > > > >>> > > > > > > > > > >>> > > > > > The PYPI stats show only downloads not current number > of > > > > >>> > > installations > > > > >>> > > > - > > > > >>> > > > > ( > > > > >>> > > > > > https://pypistats.org/packages/apache-airflow), but I > > > > >think they > > > > >>> > are > > > > >>> > > > > kind > > > > >>> > > > > > of indicative numbers for potential users of 2.0. > > > > >>> > > > > > They show steady decline of 3.5 downloads - currently > at > > > > >below 5%. > > > > >>> > > > Python > > > > >>> > > > > > 3.6 is by far the most popular (around 40% of > downloads). > > > > >>> > > > > > > > > > >>> > > > > > There are few things we could benefit from by switching > > to > > > > >3.6 - > > > > >>> > this > > > > >>> > > > is > > > > >>> > > > > my > > > > >>> > > > > > (subjective) selection of the ones that matter for > > > > >Airflow: > > > > >>> > > > > > > > > > >>> > > > > > - Formatted string literals : > > > > >>> > > > > > > > > > >>> > > > > > > > > > >>> > > > > > > > > >>> > > > > > > > >>> > > > > > > >>> > > > > > > > > > > > > > > > > https://docs.python.org/3/whatsnew/3.6.html#pep-498-formatted-string-literals > > > > >>> > > > > > (my > > > > >>> > > > > > favourite one) > > > > >>> > > > > > - Type annotations for variables : > > > > >>> > > > > > > > > > >>> > > > > > > > > > >>> > > > > > > > > >>> > > > > > > > >>> > > > > > > >>> > > > > > > > > > > > > > > > > https://docs.python.org/3/whatsnew/3.6.html#pep-526-syntax-for-variable-annotations > > > > >>> > > > > > - > > > > >>> > > > > > now especially that we use MyPy more and more this one > is > > > > >the > > > > >>> > only > > > > >>> > > > > > place we > > > > >>> > > > > > have to leave ugly comments rather than annotations. > > > > >>> > > > > > - Path-like types: > > > > >>> > > > > > > > > > >>> > > > > > > > > > >>> > > > > > > > > >>> > > > > > > > >>> > > > > > > >>> > > > > > > > > > > > > > > > > https://docs.python.org/3/whatsnew/3.6.html#pep-519-adding-a-file-system-path-protocol > > > > >>> > > > > > - > > > > >>> > > > > > nicer handling of file access > > > > >>> > > > > > - Local Time disambiguation: > > > > >>> > > > > > > > > > >>> > > > > > > > > > >>> > > > > > > > > >>> > > > > > > > >>> > > > > > > >>> > > > > > > > > > > > > > > > > https://docs.python.org/3/whatsnew/3.6.html#pep-495-local-time-disambiguation > > > > >>> > > > > > - > > > > >>> > > > > > this had already caused problems in the past (I fixed a > > > > >problem > > > > >>> > > > where > > > > >>> > > > > > pendulum and datetime objects were mixed and caused > wrong > > > > >>> > > behaviour > > > > >>> > > > on > > > > >>> > > > > > 3.5) > > > > >>> > > > > > - Json loads supports binary format - > > > > >>> > > > > > https://docs.python.org/3/whatsnew/3.6.html#json - > this > > > > >has > > > > >>> > > already > > > > >>> > > > > > bitten us as well. there was code working fine in py2.7 > > > > >and 3.6 > > > > >>> > > but > > > > >>> > > > > not > > > > >>> > > > > > working with 3.5(!). > > > > >>> > > > > > > > > > >>> > > > > > Last but not least - it might free some resources on > > > > >Travis (I hope > > > > >>> > > > > GitLab > > > > >>> > > > > > fix will be out in 10 days or so and we will be able to > > > > >start > > > > >>> > testing > > > > >>> > > > > > migration to it). > > > > >>> > > > > > > > > > >>> > > > > > J.. > > > > >>> > > > > > > > > > >>> > > > > > On Tue, Nov 12, 2019 at 10:00 PM Bolke de Bruin > > > > ><[email protected] > > > > >>> > > > > > > >>> > > > > wrote: > > > > >>> > > > > > > > > > >>> > > > > > > Hi All, > > > > >>> > > > > > > > > > > >>> > > > > > > Can we drop python 3.5 support and switch to 3.6 as a > > > > >minimum? > > > > >>> > > > > > > > > > > >>> > > > > > > Cheers > > > > >>> > > > > > > Bolke > > > > >>> > > > > > > > > > > >>> > > > > > > > > > >>> > > > > > > > > > >>> > > > > > -- > > > > >>> > > > > > > > > > >>> > > > > > Jarek Potiuk > > > > >>> > > > > > Polidea <https://www.polidea.com/> | Principal > Software > > > > >Engineer > > > > >>> > > > > > > > > > >>> > > > > > M: +48 660 796 129 <+48660796129> > > > > >>> > > > > > [image: Polidea] <https://www.polidea.com/> > > > > >>> > > > > > > > > > >>> > > > > > > > > >>> > > > > > > > >>> > > > > > > >>> > > > > > > >>> > > -- > > > > >>> > > > > > > >>> > > Jarek Potiuk > > > > >>> > > Polidea <https://www.polidea.com/> | Principal Software > > Engineer > > > > >>> > > > > > > >>> > > M: +48 660 796 129 <+48660796129> > > > > >>> > > [image: Polidea] <https://www.polidea.com/> > > > > >>> > > > > > > >>> > > > > > > > > > > > > > > -- > > > > Jarek Potiuk > > Polidea <https://www.polidea.com/> | Principal Software Engineer > > > > M: +48 660 796 129 <+48660796129> > > [image: Polidea] <https://www.polidea.com/> > > >
