Why travis failed in master branch, but successed in PR

I think this is a bug, because all the configuration is same in PR and master branch. It successed in PR, but failed in master branch without useful infomation:

the PR job is: https://travis-ci.org/github/SpaceVim/SpaceVim/jobs/726594076

the master job is: https://travis-ci.org/github/SpaceVim/SpaceVim/jobs/726584968

I want to know is there any differences between the master branch building and pr building?

I am highly skeptical that this is a bug on our end.

Your script


is setting errexit, and it is failing with the first command that exits with nonzero. It is hard for me to say what is causing the problem, but you are the best person to trouble shoot that issue.

You may also have stale cache, in which case deleting it can help you sort out the issue.

Additionally, we can turn on the debug feature if you are interested.

@BanzaiMan I’m more concerned why there’s no “<command> exited with <code>” message in the log after the failed command’s output. Looks like a case of lost output. They produce lots of output after “Writing coverage file .coverage_covimerage.”.

Regarding the OP’s issue, this does seem to be a user script error.

A workaround for losing output is to avoid producing very much output very quickly like the coverage file dump:

then you should be able to see the output from where the error happens and diagnose it.

TThanks for reply. But I do not understand why the PR building sucessed? only branch building is failed. the logic and commands are all same.

There may be different cache, there may be different set of configuration envvars, there may be external factors or code changes. Impossible to say without seeing the actual error.

Is it possiable disable all the fuck cache?