This looks reasonable to me. If everybody agrees, we should add it to the dev doc.
Thanks, Wangda On Mon, Apr 13, 2020 at 11:56 PM Wilfred Spiegelenburg <[email protected]> wrote: > Hi, > > We have been using the github squash and commit button to merge requests > for a most of our commits. I have noticed a problem with that usage. The > commit that is added by GitHub will be attributed to the person that opened > the PR (that is good and correct). However the committer is set to github. > This means that we cannot track back a commit to a committer unless you use > the github UI and open the PR. The code that is committed into the repo is > also not signed off by the person performing the commit but by using a > general github signature. > > As an example [YUNIKORN-85] shows the following commit log entry: > -+-+-+-+- > Author: Tao Yang <[email protected]> > AuthorDate: Sat Apr 11 01:55:41 2020 +0800 > Commit: GitHub <[email protected]> > CommitDate: Fri Apr 10 10:55:41 2020 -0700 > -+-+-+-+- > > And on the UI it just shows Tao committed the change while Weiwei was the > person that merged. You cannot find the correct detail unless you dig into > the original PR on github itself. > > -+-+-+-+- > [YUNIKORN-85] Improve recovery performance by querying all pods once … … > TaoYang526 committed 4 days ago > -+-+-+-+- > > Because of this I already switched back to a manual squash and commit of > the changes setting the author etc. That shows up correctly in the logs: > -+-+-+-+- > Author: Weiwei Yang <[email protected]> > AuthorDate: Fri Apr 10 02:17:16 2020 +1000 > Commit: Wilfred Spiegelenburg <[email protected]> > CommitDate: Fri Apr 10 02:17:16 2020 +1000 > -+-+-+-+- > And also in the github UI: > -+-+-+-+- > [YUNIKORN-72] data race in unit test (#96) … > yangwwei authored and wilfred-s committed 5 days ago > -+-+-+-+- > > I want to propose that we all go back to that way so we do not lose > the information of whom committed and get the correct signatures on the > committed code. > > Please let me know if this is acceptable. > > Wilfred >
