git-rebase produces incorrect output
- From
- Pavel Roskin <plroskin@gmail.com>
- Date
- Nov 29, 2019, 08:21 UTC
- Message-ID
- <CAN_72e2h2avv-U9BVBYqXVKiC+5kHy-pjejyMSD3X22uRXE39g@mail.gmail.com>
Hi!
I've discovered an issue with "git rebase" producing a subtly incorrect file. In fact, that files even compiled but failed in unit tests! That's so scary that I'm going to stop using "git rebase" for now. Fortunately, "git rebase --merge" is working correctly, so I'll use it. Too bad there is no option to use "--merge" by default.
The issue was observed in git 2.23 and reproduced in today's next branch (2.24.0.449.g4c06f74957) on up-to-date Fedora 31 x86_64.
I've created a repository that demonstrates the issue: https://github.com/proski/git-rebase-demo
The branch names should be self-explanatory. "master" is the base, "branch1" and "branch2" contain one change each. If "branch1" is rebased on top of "branch2", the result is incorrect, saved in the "rebase-bad" branch. If "git rebase -m" is used, the result is correct, saved in the "merge-good" branch.
The files in "rebase-bad" and "merge-good" have exactly the same lines but in a different order. Yet the changes on branch1 and branch2 affect non-overlapping parts of the file. There should be no doubt how the merged code should look like.
I believe the change on branch2 shifts the lines, so that the first change from branch1 is applies to a place below the intended location, and then git goes back to an earlier line to apply the next hunk. I can imagine that it would do the right thing in case of swapped blocks of code. Yet I have a real life example where it does a very wrong thing.
Indeed, "git diff origin/branch2 origin/rebase-bad" and "git diff origin/branch2 origin/merge-good" both produce diffs of 9957 bytes long, different only in the order of the hunks.
Another interesting data point - "git rebase --interactive" is working correctly.
-- Regards, Pavel Roskin