Test blurb in Travis-CI

[pitrou]

No description provided.

[brettcannon]

If #146 goes in then the .travis.yml could be turned into a matrix where blurb is tested with Python 3.5 and cherry_picker with 3.6 and an environment variable of TESTING to make it obvious which build is testing what tool.

[pitrou]

Can't cherry_picker be tested be with 3.5 as well? Or doesn't it work there?

[Mariatta]

cherry_picker doesn't work in Python 3.5.

[pitrou]

This is pending fix of issue python/blurb#1 for blurb tests to actually run.

[larryhastings]

Currently this has a conflict. I'd be happy to review it, but can you fix the conflicts first? Thanks.

[pitrou]

Ok, conflicts fixed.

[larryhastings]

There's now a merge conflict in .travis.yml. It looks like an easy merge based on a simple visual inspection, but I don't know anything about it (e.g which section should come first?) and I'm loathe to touch something I absolutely don't understand. Can someone fix the merge conflict for me?

p.s. I promise to be sliiiightly more responsive to blurb PRs in the future

[brettcannon]

I've resolved the merge conflict, now we wait if Travis agrees that I didn't muck it up. 😉

[Mariatta]

Tests failed ❌

[larryhastings]

I'll wait to (re-)review until the tests pass.

[larryhastings]

Oops! The tests didn't pass, but I hit the magic "squash and merge" button. So this work is committed!

Antoine: We're still at the PyCon sprints (last day), and we'll keep an eye on it. It seems like the Travis CI part is fine; if there are bugs, can we ping you and get another PR out of you?

[larryhastings]

Tests fixed by #246. We have Travis CI running on blurb now! Thanks, Antoine!

[webknjaz]

Such matrix + conditionals can become more messy and hard to read, so I've refactored it a bit in #252 according to best practices