Python 3 is killing Python

[Steven D'Aprano]

Terry Reedy wrote:

> On 8/6/2014 9:47 AM, beliavsky at aol.com.dmarc.invalid wrote:
> 
>> Fortran compiler vendors such as Intel, IBM, Oracle/SUN and open
> 
> *Vendors* sell compilers for money, which they can then use to *pay*
> people to do unfun stuff that volunteers don't want and should not have
> to do.

Red Hat does this, and will offer support for 2.7 until 2023. But even Red
Hat doesn't offer to support software forever -- Python 2.3 has just
reached end of life with paid Red Hat support earlier this year. Anyone
still using 2.3 now has two choices, keep using it without support, or
upgrade. The same will apply to 2.7 users. It's not like their computer
will suddenly stop running 2.7.

Come 2020 when Python 2.7 stops receiving free support from the core
developers, there's a business opportunity for the 2.7-naysayers. Red Hat
support is Red Hat Enterprise Linux only. There may be paid support from
companies like ActiveState, but that's likely to be Windows only (I could
be wrong about that). So there's the opportunity: in 2020, those naysayers
who are convinced that Python 3 is a mistake can offer paid support on
whatever platforms they like.


> Actually, I am beginning to think that 2.7 should be split off for 3.x
> development and charged for.

Python is open source. Anyone can fork it, and release 2.8, 2.9, 2.10, as
many versions they like. The only thing they can't do is call it "Python
2.8" without agreement from the PSF, which they won't get.

But they won't fork it, for two reasons: Complaining is cheap, actually
maintaining a programming language is hard work. And deep down they know
that a fork will be just a waste of time. This is not like the fork of the
X-11 windowing system a few years back, for all the complaints and whinging
about Python 3 even the nay-sayers know that the world will remain full
behind Guido, the core developers and the PSF, who are all committed to
Python 3.

Let me be frank here: the core developers are committed to making the
process of migrating from 2 to 3 as easy as possible without compromising
Python 3 in any serious manner. E.g. minor cosmetic warts, like the
re-introduction of u"" syntax just for backwards compatibility reasons, may
be allowed, reversing design decisions like strings being Unicode rather
than bytes will not be. But ultimately, people will need to make a choice:

- spend the time and effort and money to migrate from Python 2 to 3;

- spend an order of magnitude more time and effort and money to 
  re-write their applications in another language;

- pay somebody to support Python 2.7 for as long as needed;

- or do without bug fixes and security updates.

If you want bug fixes, security updates AND feature enhancements, for free,
you have have to migrate to Python 3 (or another language). If you're
unhappy with that, write to Oprah, I'm sure she'll listen.




-- 
Steven