Python 3 is killing Python

[Steven D'Aprano]

On Wed, 28 May 2014 14:58:05 -0500, Larry Martell wrote:

> On Wed, May 28, 2014 at 2:49 PM, Paul Rubin <no.email at nospam.invalid>
> wrote:
> 
>> Larry Martell <larry.martell at gmail.com> writes:
>> > Somthing I came across in my travels through the ether:
>> > [1]https://medium.com/@deliciousrobots/5d2ad703365d/
>>
>> "Python 3 can revive Python" https://medium.com/p/2a7af4788b10
>>   long HN comment thread: https://news.ycombinator.com/item?id=7801834
>>
>> "Python 3 is fine" http://sealedabstract.com/rants/python-3-is-fine/
>>
>> OT: wow that medium site is obnoxious.
>>
>>
> No company that I work for is using python 3 - they just have too much
> of an investment in a python 2 code base to switch. I'm just saying.

Is that Python 2 code base aimed at Python 2.7 or 2.6? Or 2.5? Or 2.4? Or 
even 2.3? Or all of the above?

One of the most pernicious myths about this is that there is one single 
Python 2 ecosystem. There isn't. The company I work for is stuck with 2.6 
for the foreseeable future, because that's the version of Python provided 
by the OS of choice. I recently migrated a client's code base from 2.3 to 
2.6, and they will likely stay with 2.6 forever. And I know of at least 
one company who is using Python 1.5 (yes, 1.5) and have no plans to 
migrate. 1.5 works for them, and they apparently don't need or don't care 
about security updates, so why should they migrate?

This is all good. If 2.x works for your application, and you don't care 
about all the awesome new features in 3.3+, don't care about bug fixes 
and security updates, and don't mind being stuck with a version of Python 
that will slowly but surely become more and more obsolete, more power to 
you.

The Python core developers have recent committed to providing security 
updates for 2.7 until 2020. And Redhat have paid support for 2.7 until 
2023. So there's no rush.

But anyone who makes that decision to stay with 2.x forever is in the 
same position as those who stay with 1.5 forever. Eventually, you'll have 
no OS support, no vendor support, no security updates, no bug fixes, it 
will become harder and harder to find programmers who know that 
particular version of the language, and even harder to find third party 
libraries that support it, training new staff in the obsolete version 
will be hard because all the books and tutorials will be written for more 
recent versions...

My prediction is:

- over the next three or four years, there will be a steady trickle of 
people complaining about Python 3, slowly fading as more people move to 
Python 3;

- when the main Linux distros start using Python 3 as their system 
Python, there will be a sudden rush of people to Python 3;

- about six months before Python 2 drops out of free support, there will 
be a sudden flood of panicky cries for help from people who didn't bother 
making a *single* step towards migration over the previous ten years, and 
suddenly realise that they need to migrate yesterday.



-- 
Steven