Adding python 3.7-dev to the travis matrix - #7416

Hi there @bsipocz 👋 - thanks for the pull request! I'm just a friendly 🤖 that checks for issues related to the changelog and making sure that this pull request is milestoned and labeled correctly. This is mainly intended for the maintainers, so if you are not a maintainer you can ignore this, and a maintainer will let you know if any action is required on your part 😃.

I noticed the following issue with this pull request:

• Changelog entry not present (or pull request number missing) and neither the Affects-dev nor the no-changelog-entry-needed label are set

Would it be possible to fix this? Thanks!

If there are any issues with this message, please report them here.

Looks like h5py is not yet compatible. Shall we run tests with fewer dependencies?

@saimn - Yes, I'm removing any that is not yet available.

It seems that many packages are not yet ready:

Failed building wheel for pyyaml
Failed building wheel for pandas
Failed building wheel for bintrees
Failed building wheel for scikit-image
Failed building wheel for scipy
Failed cleaning build dir for scipy

@saimn @pllim - I'm not sure how to proceed here, put it into allowed to fail and wait for fixes in follow-up PR, or do those PR against this one and merge once everything passes?

IMHO we can just wait until conda (and others) catches up. I doubt everyone is going to suddenly upgrade to Python 3.7 the day it comes out. I would even be so bold to propose a 2-week window after Python 3.7 release to do any remaining compatibility work on our side (they would be high priority then, of course). What do you think?

Maybe we can wait until the first release candidate ? This should come at the end of may: https://www.python.org/dev/peps/pep-0537/#id5

OK, then I'll just leave this here and and update as required when the RC is out.

Python 3.7.0rc1 is available, and final version scheduled to 2018-06-27, so it would be good to test in the coming weeks. Not sure if it is available with conda though.

Thanks for the update @saimn . FWIW I restarted the Travis CI build here. 🤞

Still failed. FYI https://travis-ci.org/astropy/astropy/jobs/373754558

Some of the failures we see look the same that @saimn has reported upstream (cython/cython#2043) and that has hopefully been addressed in the latest cython release. (Failures were seen with 0.27.3)

Your pinning didn't work? From https://travis-ci.org/astropy/astropy/jobs/392732161

Cython: 0.27.3

Python 3.7 was released today. FYI

(Sorry, press wrong button, hence the accidental closing.)

@bsipocz - the conflict is because of the numpy>=1.13 merge, which adjusted .travis.yml. Possibly a good excuse to double-check that you like the changes I made.

I was about to rebase and push it back out here.

Update: I changed my mind and opened #7607 instead. I went the lazy route and squashed commits and moved things around, so I'd rather we have multiple PRs than losing history here.

My attempt over at #7607 did not get very far. It looks like we might have to give the world a few weeks to make packages available here and there.

You mean numpy/numpy#11418?

@bsipocz - #7617 should have fixed at least the astropy-related problems, so maybe worth reviving this one! Note that I removed the no-changelog-needed label since somewhere we should mention that we now support 3.7, and this PR seemed as good as any - though I'm quite happy to have a new issue for that if that works better.

Sadly Travis support for Python 3.7 is not ready yet: travis-ci/travis-ci#9815
Any idea for the conda side ? I think I read that they now have Python 3.7, though the scientific packages are maybe not yet ready.

closing in favour of #7607