Closed jrochkind closed 5 years ago
I think real issue may be this one:
Travis changed the underlying platform/OS setup you are running on -- or may be has a bug which makes it unpredictable what platform will be used if you don't specify it. We were perhaps using trusty
before, and putting dist: trusty
in may make it use the same one that was working before again, and make it work again. Let's see.
Aha, yes maybe. https://changelog.travis-ci.com/xenial-as-the-default-build-environment-99476
So default distro on travis changed sometime in the past few months, gradual rollout, we haven't been committing to chf_sufia much lately, guess it happened sometime after [last succesful PR])(https://github.com/sciencehistory/chf-sufia/pull/1280) July 29.
Looks like there were a couple PR's in August that were merged by @eddierubeiz despite failed travis (cause they were simple enough changes to be confident anyway), failing for same reason.
So ok at least we understand what happened now! And putting dist: trusty
in our .travis.yml should make builds work again.
Try get around weird rubygems bug that's suddenly started affecting us: https://bundler.io/blog/2019/05/14/solutions-for-cant-find-gem-bundler-with-executable-bundle.html
Looks like this in travis log with test failure:
Not sure why it doesn't reproduce locally, just on travis.