Add support for duet-rebase and duet-cherry-pick and git stash

[thewoolleyman]

Hi,

I use a rebase workflow for feature branches. Are you open to adding a duet-rebase command? If so I can try to submit a pull request if you don't get to it.

Thanks,
-- Chad

[jszwedko]

@thewoolleyman definitely open to a pull request!

[thewoolleyman]

@jszwedko Hi - I'm hacking on this today and having some troubles getting my env/tests green before I start. Would you be available to answer a couple of questions? I've got bats and getdb.io installed (I think correctly) but still have about half of the tests in the suite failing...

(update: and go install ./... fails on the "internal" dependencies)

[jszwedko]

@thewoolleyman interesting, could you share the failing tests and their outputs? I'm on Freenode as _jesse_ if you think it'd be useful to communicate in real time.

I appreciate you contributing!

[thewoolleyman]

@jszwedko Sure - it would probably be easiest to talk through real time. What's a good time to discuss on freenode?

[jszwedko]

@thewoolleyman awesome, I should be online most of today (PST hours) or tomorrow during 9AM - 5PM PST.

[thewoolleyman]

OK, Pinged you on freenode. Here's a gist with info on where I am including details of my env and failing tests: https://gist.github.com/thewoolleyman/4fdf9529e8fef5bbbf5a - I'm still in private chat on freenode to discuss further.

[thewoolleyman]

OK, after brew uninstall git-duet and manually adding absolute path to project bin (after gb build) the tests passed. Thought I had tried this before, but I guess not :)

I updated the gist for completeness, and I'll look at adding tests after we resolve exactly how you want git-duet-rebase implemented (per discussion on the pull request: #25 (comment) )

[thewoolleyman]

@jszwedko I also remembered git stash was the other one that always fails. Any reason not to also add an executable wrapper for this too?

[krishicks]

The way I've been handling this is to use git rebase -i --exec "git duet-commit --amend --reset-author". It's a bit annoying because you still need to remove the extra signoff line in each commit, but at least it fixes the author/committership.

[jszwedko]

That is a good one line work around, thank you @krishicks -- I'll add that to the README until we have a better way to handle this (it is annoying that the signoff line will be inaccurate at that point).

[jszwedko]

Added workaround to the README in bebab41

Hopefully we can come up with something better for the long term.

[krishicks]

Cool. You might want to specify in the README that you still need to update the signoff by deleting all but the last entry.

For what it's worth, the git hooks samples have one that adds Signed-off-by (prepare-commit-msg.sample), which includes this:

# The third example adds a Signed-off-by line to the message, that can
# still be edited.  This is rarely a good idea.

I think it's that way for this very reason of adding multiple Signed-off-by entries.

[carpeliam]

@pivotaljohn just shared https://gist.github.com/pivotaljohn/8bd5dd88cc98e1c5381d5128d81828a1#file-git-duet-rebase-sh with me... thoughts?

[krishicks]

Seems a bit much. filter-branch is the nuclear option.

[jszwedko]

filter-branch is very powerful, but seems reasonable for this use case. I'll leave the interactive rebase as the preferred option (since you are less likely to do any foot shooting), but I'll add this as an additional option. Thanks for sharing!

[jszwedko]

Added script and link in cea5358