OpenMRS Core Master CI build is broken (we can do better!)

@dkayiwa, @wyclif,

The OpenMRS Core Master build is broken on CI. It started breaking 16 hours ago, I guess due to merging a PR from Wyclif here: Daniel, I see that you then tried several manual builds (that still failed) and left a comment on the PR asking Wyclif if he knew why it was failing.

But that’s the wrong workflow, because now it’s end-of-day Friday and we’ve left our master CI build broken over the weekend. The commit should have been reverted.

In fact, didn’t we just have a long discussion about this topic a couple months ago?

I just went ahead and reverted the PR (it was just a few clicks in github, starting from the merged PR):

(I’m heading to the airport now, so hopefully this does fix the CI build…)

It may feel better to leave “mostly-working” code in, but this is wrong. We need to revert.


FYI, I heard that @wyclif will be on holiday for the next 20 days, so if you’re waiting on him, you may be waiting a long time…

In @dkayiwa’s defense. Travis CI passed for this PR – so there was really no indication that the build would fail…though breaking the build is never a good thing…just pointing that out…but there are cases where the tests will randomly fail…

Is there way that we can add a a check for Bamboo with each PR?

Just to be specific, the problem isn’t that the build broke: that will happen sometimes.

The bad habit that we must break, as a community, is leaving the build broken if you’re not actively debugging it. (Especially we shouldn’t be leaving builds broken overnight or over the weekend.)

-Darius (by phone)