In this PR codecov produce the wrong report. The report includes files that are not part of commits and so coverage report is wrong. It is happening for while now in all pull requests. It seems compared commits are correct but the report is wrong. Codecov is run by Travis. What are we doing wrong?
Codecov is running on Travis: Ubuntu 16.04.6 LTS, codecov version: 2.0.22
Hey @PrimozGodec – sorry for the lengthy delay. I agree that there was only one file changes in this PR and yet Codecov picked up 4-5 changed files, which is incorrect.
Almost certainly this is related to Codecov picking the wrong parent/base commit to compare against.
Do you have a more recent example of this happening again? We lost the logs on this one, sorry 
When you send it, can you send the commit SHA in question and the commit SHA of the parent you would expect from the git tree?
Thank you for considering this issue.
It is happening in practically all Pull requests at this repository.
For example this one: PythagoreanTree: data info displayed in the status bar by aturanjanin · Pull Request #4699 · biolab/orange3 · GitHub
Commit SHA is: a5f3a775b9e3355f1d2a1e040e42bda2e1f020b2
Parents commit: cf0f69fe9535c02e71a56861edf05ae573e12cf0
Right, so we can actually see that indeed although the head commit is correct, the parent is wrong:
And moreover, the base commit of the PR is being used at the parent commit of the head commit:
That base commit “CI Failed”, you can see the builds here: Codecov
What is the relationship of the the parent commit that you stated and this head commit? It seems to be on a different PR:
Also, it seems like that CI Failed:
Can you check why CI Failed on both the actual parent commit and the parent commit that Codecov picked?
Hi @PrimozGodec, just wanted to check back in on this and see if you were still having issues.
Hi thank you for ping. We had some other issues with CI so we couldn’t not check what is happening with codecov. I will go back to this project and will let you know. I also tried one other thik which is potential source of the problem.
Hi @PrimozGodec, going to close this out for now, but feel free to open a new thread in #support if the issue comes back up
