Let's drop support for Python 2.6

[whart222]

I don't see any meaningful reason to continue support for Python2.6, but let's start a conversation here about this possible change.

I've seen evidence that we're spending time supporting this version of Python without deriving any meaningful utility.

[hugovk]

Here's the pip installs for pyomo from PyPI for the last month (via pypinfo --percent --pip --markdown pyomo pyversion) -- no Python 2.6.

python_version	percent	download_count
2.7	54.4%	1,141
3.5	20.6%	433
3.6	20.1%	421
3.4	4.7%	99
3.7	0.1%	3

[whart222]

@hugovk Nice! I didn't realize that would could get that info from that data store.

[hugovk]

Yeah, you can get all sorts of things out of it, see https://github.com/ofek/pypinfo for some other interesting examples.

[ghackebeil]

I'm trying to think of things that we can start doing in code once we officially drop support for 2.6 (hoping others will add to the list). A few things that come to mind:

1. Alternative (and much faster) dict initialization syntax: {key:1.5 for key in range(10)}
2. Instead of nesting with blocks one can simply separate with arguments with a comma, which allows reducing the number of indentation levels needed (e.g., with A() as a, B() as b:)

[hugovk]

Reasons for dropping Python 2.6

General

• Unsupported since 2013-10-29 and no longer receiving security updates from the core Python team
• https://snarky.ca/stop-using-python-2-6/
• http://www.curiousefficiency.org/posts/2015/04/stop-supporting-python26.html
• Python 2.7 will be going away in the not too distant future, some projects are already planning for dropping 2.x. Dropping 2.6 sooner may make dropping 2.7 easier http://www.python3statement.org
• Current pip 9 deprecates Python 2.6 support, pip 10 won't support it (Drop support for Python 2.6? pypa/pip#3955)
• Wheel no longer supports Python 2.6 pypa/wheel@c9e5486
• Not much PyPI traffic (June 2016) Deprecate/Drop Support for Python 3.3? pypa/pip#3796

Code

• Can use automatic formatters eg. "{} {}".format(a, b) rather than "{0} {1}".format(a, b) https://pyformat.info
• Can drop dependencies on argparse, unittest2 and ordereddict
• Can remove quite a few sys.version_info and sys.hexversion and code branches
• Can drop a ~10 minute build job from an already quite slow Travis CI test matrix
• Can update list, dict, set and tuple handling, like this https://github.com/adamchainz/flake8-comprehensions#examples