I [talked about this on Slack](https://cordova.slack.com/archives/C068CHRJ5/p1535978572000100). The nightly build are sometimes broken.
I believe that [the `readFile` call](https://github.com/apache/cordova-coho/blob/5d710a3bb1838d1e822ee0ed4f830374c6c4b180/src/versionutil.js#L134) is invoked and then [the `gitutil.pendingChangesExist()` call is made](https://github.com/apache/cordova-coho/blob/5d710a3bb1838d1e822ee0ed4f830374c6c4b180/src/versionutil.js#L166) right after. Both calls have asynchronous behavior, which introduces a race condition. I believe that this race condition is to blame for the random build failures. Looking around the file, there is [at least another case](https://github.com/apache/cordova-coho/blob/5d710a3bb1838d1e822ee0ed4f830374c6c4b180/src/versionutil.js#L134) of this pattern. There might be more in the module if this is a general oversight by the author. My suggestion is, given the generator/yield approach of the module, the workflow should be adapted to that or, for a smaller refactoring, the code should use `fs.readFileSync` instead. In the latter case, refactoring the code should be trivial. I could create a PR for that if there is consensus about the change. [ Full content available at: https://github.com/apache/cordova-coho/issues/195 ] This message was relayed via gitbox.apache.org for [email protected]
