Git question.md

Hello and thanks for reading,

I've a Redis pull request that no longer applies because, for example, one line where there was a < b is now a > b, otherwise the patch would apply cleanly.

if I try to cherry pick the commit and resolve the conflicts, it works, it's just a lot of useless work editing the conflicting code to merge properly and there is even the risk of making mistakes.

An alternative is to:

1. Modify back the source code to a < b with a commit called FOO.
2. Apply the pull request commit cleanly with cherry pick.
3. Revert to a > b and amend the commit to make the change.

However this way I've the spurious commit FOO. If I rebase with rebase -i and squash the commit, the PR author name is removed from the history, which is very unfortunate. Is there a better alternative? Thanks in advance.

It's not super elegant, but maybe you can use the GIT_AUTHOR_X env variables to set the authorship?

Starting with the series of commits you've described (revert, cherry pick, revert-the-revert), I think this does it:

git checkout pr
git reset --soft master
GIT_AUTHOR_NAME='Joe Smith' GIT_AUTHOR_EMAIL='joe@example.com' GIT_AUTHOR_DATE='2016-04-28 07:00:00 -0700' git commit -m 'original commit message'

This leaves the pr branch with one commit in it, and it's marked as being authored by the original committer.