Re: odd behavior with git-rebase

[Jeff King <peff@peff.net>]

On Mon, Mar 26, 2012 at 02:31:05PM -0400, Phil Hord wrote:

> Coincidentally I ran into this same behavior this week.  But what
> bothered me about it was the messages git gave me.  The empty commit
> gave me cherry-pick hints instead of rebase ones, including advising
> me to "use 'git reset'" to resolve the problem if I don't want this
> commit after all.
> 
> $ git rebase -i HEAD~10
> ...
> The previous cherry-pick is now empty, possibly due to conflict resolution.
> If you wish to commit it anyway, use:
> 
>     git commit --allow-empty
> 
> Otherwise, please use 'git reset'
> # Not currently on any branch.
> nothing to commit (working directory clean)
> Could not apply d513504... Some commit message
> 
> 
> I'm not sure if this is the norm or if it's a result of some other
> things I did in this sequence.  But I've seen it several times now.
> I've only tested it on 1.7.10 versions, including RC2.

This is easily reproducible on a simple test case:

  commit() {
    echo $1 >$1 && git add $1 && git commit -m $1 && git tag $1
  }

  git init repo &&
  cd repo &&
  commit one &&
  commit two &&
  git commit --allow-empty -m empty &&
  commit three &&
  git checkout -b fork one &&
  commit four &&
  git rebase -i fork master
  git --no-pager log --oneline

(this is the same test case I used without "-i" to check the rebase
skipping behavior).

I agree the mention of cherry-pick is a little confusing. I think the
advice to use "git commit --allow-empty" is still the right thing
(although better still would be to recognize that the commit was empty
in the first place and not stop at all). I think the message is showing
the fact that "rebase -i" is cobbled together from other pieces. I
wonder if the sequencer work would make this a little smoother (I
confess I have not paid much attention to what is happening in that
area).

-Peff