Hi, Piotr, Thanks for posting this for discussion.
In my work projects we use Python 3.6, 3.7, and (cough) 2.7. So it won’t hurt me directly if Avro doesn’t support Python 3.4 anymore. However, avro-py3 is a library, so wide version support is appropriate in my opinion. Python still supports 3.4. There will be one more supported release next March if it all goes according to the release schedule in PEP 429. As you previously pointed out, 13% of pypi downloads in one recent month indicated 3.4. But I don’t think that’s a small number. I think that’s a worthwhile-sized minority. I would love to use fstrings. Even moreso I want to use native type hints, and then get mypy running on yetus. But I genuinely think supporting 3.4 as long as Python does is the right thing to do for now. That said, I think we should also plan to deprecate it soon. We can drop deprecation warnings in the code now and plan to obsolete it completely in 2019. That’s my opinion. If Apache/avro has an official version compatibility support policy, someone please clue us in! https://www.python.org/dev/peps/pep-0429/#id4 On Thu, Nov 8, 2018 at 14:28 Piotr Gołąb <[email protected]> wrote: > Hi Michael and Fokko, > > I posted some notes today morning, but it was too late I guess. > > Isn't the decision to use Python 3.4 too conservative? 3.4 was released on > 2014 and I strongly believe that we should be as up-to-date as possible. > 3.4 > is used by around 13% of users as shown by Scott Belden and indeed this > will > shrink. At the same time we cannot use some of the features that seem > obvious in Python 3 (f string formatting, etc.). > > Also I think supporting three last minor versions is a good practice, for > example pandas is supporting Python 3.5, 3.6 and 3.7. What do you think > about this? > > By the way, big thanks for supporting the Python Avro! I'm really glad that > you're bringing up this discussion. > > Best, > Piotr (sireliah) > > > > > > -- > Sent from: > http://apache-avro.679487.n3.nabble.com/Avro-Developers-f679485.html >
