Test multiple versions of k8s

[foot]

• First go, use even more yaml anchor helpers

Fixes #288

[dlespiau]

19 checks successful! 😱

looks great to me ⭐ :)

I've been toying with the idea of generating travis/circle CI files from js/ts and jk to go beyond anchors, but it seems anchors can be made quite readable :)

[foot]

I've been toying with the idea of generating travis/circle CI files from js/ts and jk to go beyond anchors, but it seems anchors can be made quite readable :)

Yes! That was plan B when anchors obviously failed, but they worked okay in the end.

[rade]

This is awesome!

Two niggles

• missing trailing newline
• it would have been good to squash some of the commits, to elide missteps

[foot]

missing trailing newline

Will fix

it would have been good to squash some of the commits, to elide missteps

I tested out the "squash and merge" feature of GitHub for the first time which has made a history that looks like this now:

* baf30e1 (HEAD -> master, origin/master, origin/HEAD) Test multiple versions of k8s (#291)
| * 50102a4 (origin/288-update-integration-tests-to, 288-update-integration-tests-to) Re-run kubernetes jobs once a day in case of new k8s
| * 7c6f178 Getting bash syntax in order
| * ca03827 Add a third "stable" target for k8s testing
| * 5e9a2d7 Factor out some more things
| * c54a33b even edgeier?
| * f57578b no dice, newer machine image
| * 122583d Internet suggested hack to get minikube+circleci
| * 6c545c0 Test earlier minikube
| * 0f2c5f6 Ooops, wrong syntax, disable some jobs for now
| * 81439ba Test multiple versions of k8s
|/
*   e51f65d Merge pull request #289 from weaveworks/fixes-minikube-dns-failures

E.g. It created a new squashed commit on master rather than a merge commit. I'll avoid that in the future as its a bit confusing.

[rade]

I would stay away from "squash and merge" and be more selective. Squashing all commits into one is rarely the right level of granularity.