I do not know how to approach this issue. I will try to tell everything as it is.

I have a lot of open source projects on github. For their testing and analysis, I use such projects as:

  • Scrutinizer
  • SensioLabsInsight
  • Coveralls
  • Styleci
  • Travisci

Recently I noticed that if I run an analysis of the code on StyleCI and create a PR with corrections to the code, then the PR, after assembly, closes automatically and the branch is deleted. PR closes naturally on my behalf, which is not very pleasant. We have to restore the branch, rediscover the PR and merjit it to the master. An example of such a PR .

Small studies have shown that the PR build has collapsed due to code coverage .

Coverage decreased (-0.03%) 

Googling a bit, I stumbled upon the next ticket in the Codesracks repository. The developers of SonataAdminBundle faced a similar problem.

It turns out the problem is clearly in the service Сoveralls, but I do not understand something else. Why is PR closed? SonataAdminBundle did not close the same PR . Yes, there is an error. But nothing was done automatically with PR.

Perhaps this is somehow connected with the name of PR:

 Apply fixes from StyleCI 

Because of the word fixes GitHub could have tried to do something with an automatic machine. If there were no errors, then everything is OK , but this does not explain why PR was closed.

A small googling showed that this is not a frequent problem, but there are similar examples 1 , 2 .

Can anyone have any idea where to dig?

    1 answer 1

    can dig here:

    It seems that:

    • Coveralls in any case removes the branch (even fail, even OK)
    • and github automatically deletes the PR for a non-existing branch
    • no, if everything is OK, then the branch is not deleted. In this example, nothing happened with the gun. I myself smerdzhil PR and he deleted the branch - ghost404