Question

I'm doing a long git rebase with a lot of commits. I accidentally --skipped a commit where there were some conflicts which I resolved. I should have done git rebase --continue.

Is there are way to re-apply this previous commit during this rebase phase and then continue the rebase?

One way I see is to

  1. stop the rebase at this point by creating a branch on the last commit which was correctly applied
  2. restart the rebase starting with the previously skipped commit.

Or can I do a cherry-pick whilst being in a rebase-phase?

Was it helpful?

Solution 2

I found a way which "worked for me":

During a rebase lots of things are happening in the .git/rebase-apply directory. Amongst others there is a file called next. next is containing a number which corresponds to a file which is residing in the .git/rebase-apply as well. This file contains information about the commit which is currently being processed. For example:

$ cat .git/rebase-apply/next
0260
$ less .git/rebase-apply/0260
<info about the commit which is currently processed (and has conflicts)

Git seems to keep the commits which are skipped as the above mentioned files. Whereas files corresponding to commits which have been applied are not there anymore. The commit I accidentally skipped was called 0259 and the file was still present.

Here is what I did:

$ echo "0258" > .git/rebase-apply/next

With that I told git that currently the 258th commit is processed (which previous applied correctly). Then I did

$ git rebase --skip

to tell git to forget this one and, voila, I could again work on the skipped commit, correct the conflicts and --continue. It has worked.

OTHER TIPS

Git is great because it saves a log of basically everything you commit.

  1. Find your commit in ".git/logs/HEAD" and open in a text editor

  2. Find your SHA in the HEAD file

    3c8c... 2260dc... Full Name {my.email@email.com} 1471276956 -0600 commit: Saving Trial 1,2,3

  3. Type (note type enough of the sha so git knows which one to pull):

    git checkout -b recovery 2260d...

See link for reference: http://blog.screensteps.com/recovering-from-a-disastrous-git-rebase-mistake

The existing answers are great, but I found an even quicker way to recover the single commit I had just rebase-skipped/'lost':

  1. scroll up through your terminal output where you just made that commit
  2. copy the commit hash,
  3. run git cherry-pick -x <commit-hash> on a fresh branch.

For me, I found the rebase hash in .git/REBASE_HEAD and copied the hash. Then in my branch where I did the git rebase --skip, I did a git merge {copied hash}. And it worked!

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top