Hi @ljharb, would it be possible to run the bash command as a separate line item? Codecov typically produces an output which shows some debugging steps that are missing in the travis builds.
I’m not sure what output would be different that way - bash <(curl -s https://codecov.io/bash) -f coverage/*.json shouldn’t provide any different output being behind a &&, i think.
hm, i’ve never seen that output anywhere before. Looking at https://codecov.io/bash, all the output is via the say function which just echos to stdout, so the only things i can think of is either “travis-ci is doing something weird to suppress the output”, or, “the thing that’s failing is triggering the -e or the pipefail and exiting before there’s any output in the first place”. Not sure how to debug further.
@ljharb, apologies, it looks like you don’t have that output on test runs unless COVERAGE=true, and that is all well.
Again, this is, without a doubt, a permissions issue with the integration. I would advise going here and checking that you have the permissions set properly on your repositories and organization. You should be seeing
@ljharb, I’m really sorry I don’t have a good answer for you here at all. I’m going to guess that the integration via GitHub got corrupted somehow. Would you be able to reinstall the app?
Great, I’m so glad that worked for you. Sorry it took so long to resolve. As for a refresh, thanks for the feedback! Clearly this would have been easier to make self-service, and I will make a note to the product team.
What sort of permissions issue? You can see in the PR timeline that i’ve already force pushed it twice. I can add a commit, but i’d still need to rebase it back out and force push it before I could land the PR.
@ljharb, I’m sorry this has gone on for so long for you. I’ve made an issue with the product team, but we may need you to push up a new commit next week if you don’t mind. Thanks for your patience as we try to investigate the issue.
@ljharb, got it. We will likely need a new commit when that happens. If it’s working for you now, then maybe it was a fix we had indirectly made. If it happens again, please post here.