Multiple pull-request authors adding to AUTHORS creates merge conflicts

[kemitchell]

Hat tip: @bkeepers

[kemitchell]

@mlinksva, the merge conflicts thinkathon is here.

[kemitchell]

I don't suppose there's any way to have GitHub to apply the union built-in merge driver per a .gitattributes file?

union
Run 3-way file level merge for text files, but take lines from both versions, instead of leaving conflict markers. This tends to leave the added lines in the resulting file in random order and the user should verify the result. Do not use this if you do not understand the implications.

https://git-scm.com/docs/gitattributes

[kemitchell]

https://github.com/berneout/union-merge-authors has an example repo with:

1. master branch with existing contributor lines

2. first-pr and second-pr branches adding one new contributor each to the bottom of AUTHORS.

3. .gitattributes file configuring Git to use the union merge strategy for AUTHORS

4. Resulting, conflict-free AUTHORS file from running:

   git checkout master
   git checkout -b merge-at-command-line
   git merge --no-ff second-pr
   git merge --no-ff first-pr
   git push origin merge-at-command-line

   Which yields:

   # [AUTHORS Certificate text omitted]
   Anna Maintainer <[email protected]>
   Bob Contributor <[email protected]>
   Danielle Contributor <[email protected]>
   Charlie Contributor <[email protected]>
   

[kemitchell]

berneout/union-merge-authors#2 is a pull request showing GitHub does not apply configuration in .gitattributes to determine if there are conflicts. Bummer.

[kemitchell]

Looking into writing a bot that:

1. Users can add as a collaborator
2. Merges PRs with CLI Git on request

[kemitchell]

Related: GitLab reduced merge conflicts by 90% with changelog placeholders. Long story short, GitLab were having merge conflict problems with CHANGELOG. The solved initially by adding lots of blank placeholders, and asking people to replace a random one. They eventually went to merge=union, which I mentioned earlier.

[kemitchell]

Related: isaacs/github#487

[mlinksva]

Naive quest-suggestion: wouldn't adding authors in name or email sorted order rather than to the bottom result in fewer conflicts?