Drop support for Python 2.6

[jdufresne]

I would humbly like to suggest icalendar drop support for Python 2.6.

The last release of Python 2.6 was 2013-10-29, almost 4 years ago. It is no longer receiving security or bug fixes.

https://www.python.org/dev/peps/pep-0361/

The pip project itself has recently dropped support for 2.6. At the beginning of the year, their numbers estimated that Python 2.6 accounts for ~2% of their downloads.

pypa/pip#4343

For projects that still use Python 2.6, they can continue to pip install an older version.

I've tried my best to remove as much 2.6 specific code as I can, including the 'Programming Language :: Python :: 2.6' trove classifier from setup.py. I've also removed Travis CI testing, which should result
in slightly faster testing and fewer wasted resources.

Changed:

• Update trove classifiers, Travis test matrix, and tox test matrix
• Use set literals
• Use dict comprehension
• Remove dependencies on ordereddict and unittest2
• Remove u string prefix, use more modern unicode literals instead

[untitaker]

@thet @regebro Any opinions?

[jdufresne]

Looks like I introduced some Travis failures. I'll need to investigate those before this is merge ready.

[jdufresne]

All tests now pass.

[hugovk]

Here's the pip installs for icalendar from PyPI for the last month:

$ pypinfo --percent --pip icalendar pyversion
python_version percent download_count
-------------- ------- --------------
2.7              73.5%         28,310
3.4               9.2%          3,528
3.6               8.7%          3,351
3.5               8.4%          3,246
3.7               0.2%             59
3.3               0.1%             28
2.6               0.0%             15
3.2               0.0%              3
None              0.0%              2

You could also consider dropping 3.3, which went EOL last month and like 2.6 also has low numbers.

And here's pip installs across PyPI showing 2.6 is now more like ~1% than ~2%.

$ pypinfo --percent --pip "" pyversion
python_version percent download_count
-------------- ------- --------------
2.7              77.0%    495,806,602
3.6               9.6%     61,584,928
3.5               7.9%     50,588,704
3.4               4.1%     26,641,353
2.6               1.2%      7,944,652
3.3               0.1%        761,138
3.7               0.1%        392,332
3.2               0.0%         50,760
None              0.0%         21,424
2.8               0.0%            206
3.1               0.0%            144

[jdufresne]

I'm happy to also drop support for Python 3.3 if it is agreeable with the project maintainers. @untitaker what do you think?

[untitaker]

I'm specifically waiting for the feedback of the people I mentioned above. I don't have any objections myself, but want to wait for their response.

…

On Fri, Oct 27, 2017 at 01:20:07PM +0000, Jon Dufresne wrote: I'm happy to also drop support for Python 3.3 if it is agreeable with the project maintainers. @untitaker what do you think? -- You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: #246 (comment)

[thet]

@untitaker @jdufresne @hugovk
+1 for dropping 2.6 and 3.3 support.

Doing would normally include a major version increase - which implies adapting the semantic versioning scheme. But that's another discussion.

[jdufresne]

Thanks. Done! I've removed 3.3 support in the latest revision. All additional feedback is welcome.

[thet]

LGTM, also a good amound of nice code cleanups.
Tests pass, I'll merge.

@jdufresne great, thanks!

[thet]

Also, I'll bump the version to 4.0 I guess.

[polyzen]

May want to consider quoting the versions in .travis.yml again in the future: travis-ci/docs-travis-ci-com#1537 (comment)

[jdufresne]

Interesting. Thanks for that link. I wasn't aware of that gotcha.

[hugovk]

Good to know, at least we're some time away from Python 3.10 :)