[Python-ideas] A python bridge between versions

[Andrew Barnert]

On Feb 28, 2014, at 19:09, ian o <ian.team.python at gmail.com> wrote:

> However, you state that our situation, and that of the other 60% of python programmers as at end January 2014 who are constrained from using python 3 by python 2 dependencies:  "we don't mind about those developers"

You're twisting the numbers. I'm part of the 60% who have to work on Python 2.x for at least one project. But first, I have others--including shipping commercial code both on the desktop and the server--using 3.x. And second, in all but one case, the 2.x apps are 2.x not because of lacking library support, but because we have an already-working and -deployed app that's effectively in maintenance mode, and it's not worth the effort or risk to port it to 3.x. (In other words, most of my 2.x work points to the _success_ of the migration plan, not a failure.)

And you're also twisting Nick's words. He said that he cares more about getting library developers to 3.x than application developers. And he explained why that makes life better for application developers. To take that as "the Python devs don't care about application developers" is willfully misinterpreting the point.

> And of course the dependent library module developers will not update the library, because 'no one out there uses python 3.x'.  

None of the library devs I've ever dealt with has ever said that (well, not since the early 3.1 days).

Most projects that don't support 3.x today want to do so, it just isn't always the very highest thing on their priority list. Three times I've had to port a library myself rather than waiting around--but in every case, the library's dev was happy to take my code, even if it meant losing 2.5 support, reviewing large changelists, etc.

I'm sure there are some exceptions, but it's certainly not the ubiquitous case you're presenting.

There's a reason the Python 3 Wall of Shame changed it's name to Wall of Superpowers and the competing py3ksupport page stopped updating: because people have been porting important libraries to 3.x, and continue to do so.

> No wonder Tiobe index shows python dropping

And is the Python 3 migration also to blame for Python's direct competitors (Ruby, Perl, and PHP) also dropping)? Or for the general trend in Tiobe numbers to show ObjC and Windows-specific languages continually gaining on cross-platform languages?

> and in the words of those trying to persuade me to move the projects to another language 'python 3.x' came out 5 years ago and we still cannot move to it, this is going no where.

This is pure FUD. And I don't see how repeating that FUD to the Python devs is going to help you. If you want to convince these people to stay on Python, you need to start talking specifics instead of talking points. What libraries are stopping you from moving to 3.x? Does Ruby or Node.js or whatever actually have a better alternative for all of those libraries than Python 3 does? Would it really be easier to write those missing libraries yourself in a new language than to port existing Python 2.x libraries to 3.x?

For that matter, have you actually checked whether those libraries really are blocking you? I've lost count of the number of people who've claimed they can't switch to Python 3 because of some library that's been ported since 2010 or that has a 3.x-compatible drop-in replacement.