git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jeff King <peff@peff.net>
To: Orgad Shaneh <orgads@gmail.com>
Cc: git <git@vger.kernel.org>
Subject: Re: A bug in git merge
Date: Tue, 27 Mar 2018 08:49:43 -0400	[thread overview]
Message-ID: <20180327124942.GA18448@sigill.intra.peff.net> (raw)
In-Reply-To: <CAGHpTBJ-4a5+5=303_N-n3TtUg_sXt6J1cad0WCQ4Kc-=HDdSA@mail.gmail.com>

On Tue, Mar 27, 2018 at 12:53:52PM +0300, Orgad Shaneh wrote:

> If I cherry-pick a commit that added a line, then merge another commit
> which removes this line, the line remains in the file instead of being
> removed.
> 
> The following script demonstrates the bug.
> 
> file should be equivalent on both branches
> 
> git init
> seq 1 20 > file
> git add file
> git commit -m 'Initial'
> sed -i "s/^5/5\n55/" file
> git commit -a -m 'Added 55'
> git checkout -b other HEAD^
> git cherry-pick master
> git commit --amend --author 'Author <a@b.c>' --no-edit # generate a new hash
> sed -i '/55/d' file
> git commit -a -m 'Removed 55'
> git checkout master
> git merge --no-edit other
> git diff other # Should be equal

This isn't a bug; it's the expected behavior for a 3-way merge.

The merge looks only at the two final states to be merged, and the merge
base. The three states we have are:

    base: without line 55
    ours: with line 55
  theirs: without line 55

And since only one side made a change, the resolution is to take the
change in the final result. The fact that the other branch actually
manipulated the file (and how it manipulated it) isn't considered at
all.

The fact that it was a cherry-pick doesn't change that. A cherry-pick is
an application of the same changes, but it has no history-relationship
with the original commit in Git.

One could argue for or against the user experience of history as a DAG,
3-way endpoint merges, and how cherry-picks are stored, but this is all
working according to Git's design and not a bug.

-Peff

      reply	other threads:[~2018-03-27 12:49 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-03-27  9:53 A bug in git merge Orgad Shaneh
2018-03-27 12:49 ` Jeff King [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20180327124942.GA18448@sigill.intra.peff.net \
    --to=peff@peff.net \
    --cc=git@vger.kernel.org \
    --cc=orgads@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).