Removing a commit from a local branch

Removing a commit from a local branch

[Chris Packham]

Hi List,

I'm part of a development team using git. We use a maintainer model
where developers send patches/pull requests to a maintainer who
applies the patches to a local branch, decides if they're good or not
and pushes the good patches to the public repository.

What I want to do is script the removal of a bad patch so that the
maintainer identifies a patch in his local branch, sends an email to
the author telling them why their patch is being rejected then removes
the commit for that patch. Using git log a script can extract the
author email address, hash and headline of each commit. Based on that
information scripting the email is easy enough. Now I come to using
git rebase to remove the bad commit based on its hash which leads me
to my question - How do I refer to a commit based on the hash of its
parent?

Consider the following example. The maintainer has the following branch locally

  todeliver: A-B-C-D

He is happy with commits A, C and D but wants to reject B. Ideally I
want to be able to say
  git rebase --onto <parent of B> <child of B> todelvier

and get
  todeliver: A-C'-D'

I know that <parent of B> can be referred to as B~1 but what about
<child of B>? I've read through the man page for git-rev-parse and
nothing stands out as child of commit X.

Is there a better what do achieve what I'm after?

Thanks,
Chris

Re: Removing a commit from a local branch

[Jakub Narebski]

"Chris Packham" <judge.packham@gmail.com> writes:

> Consider the following example. The maintainer has the following
> branch locally
> 
>   todeliver: A-B-C-D
> 
> He is happy with commits A, C and D but wants to reject B. Ideally I
> want to be able to say
>   git rebase --onto <parent of B> <child of B> todelvier
> 
> and get
>   todeliver: A-C'-D'
> 
> I know that <parent of B> can be referred to as B~1 but what about
> <child of B>? I've read through the man page for git-rev-parse and
> nothing stands out as child of commit X.
> 
> Is there a better what do achieve what I'm after?

Yes, I think it would be easier to either use "git rebase --interactive"
(with some script taking place of EDITOR, or something), or prod Stephan
Beyer (CC-ed) to finish git-sequencer...

-- 
Jakub Narebski
Poland
ShadeHawk on #git

Re: Removing a commit from a local branch

[Björn Steinbrink]

On 2009.01.12 11:42:24 +1300, Chris Packham wrote:
> Hi List,
> 
> I'm part of a development team using git. We use a maintainer model
> where developers send patches/pull requests to a maintainer who
> applies the patches to a local branch, decides if they're good or not
> and pushes the good patches to the public repository.
> 
> What I want to do is script the removal of a bad patch so that the
> maintainer identifies a patch in his local branch, sends an email to
> the author telling them why their patch is being rejected then removes
> the commit for that patch. Using git log a script can extract the
> author email address, hash and headline of each commit. Based on that
> information scripting the email is easy enough. Now I come to using
> git rebase to remove the bad commit based on its hash which leads me
> to my question - How do I refer to a commit based on the hash of its
> parent?
> 
> Consider the following example. The maintainer has the following branch locally
> 
>   todeliver: A-B-C-D
> 
> He is happy with commits A, C and D but wants to reject B. Ideally I
> want to be able to say
>   git rebase --onto <parent of B> <child of B> todelvier

You don't want <child of B> there, just B.

git rebase --onto <onto> <upstream> <branch>

Rebases the commits from the range <upstream>..<branch>, and that
_excludes_ the commit (referenced by) <upstream>.

So:
git rebase --onto B^ B todeliver

Works on: B..todeliver == todeliver --not B
And that range contains commits C and D.

Björn

Re: Removing a commit from a local branch

[Chris Packham]

Thanks thats exactly what I was after.

On Mon, Jan 12, 2009 at 11:52 AM, Björn Steinbrink <B.Steinbrink@gmx.de> wrote:
> On 2009.01.12 11:42:24 +1300, Chris Packham wrote:
>> Hi List,
>>
>> I'm part of a development team using git. We use a maintainer model
>> where developers send patches/pull requests to a maintainer who
>> applies the patches to a local branch, decides if they're good or not
>> and pushes the good patches to the public repository.
>>
>> What I want to do is script the removal of a bad patch so that the
>> maintainer identifies a patch in his local branch, sends an email to
>> the author telling them why their patch is being rejected then removes
>> the commit for that patch. Using git log a script can extract the
>> author email address, hash and headline of each commit. Based on that
>> information scripting the email is easy enough. Now I come to using
>> git rebase to remove the bad commit based on its hash which leads me
>> to my question - How do I refer to a commit based on the hash of its
>> parent?
>>
>> Consider the following example. The maintainer has the following branch locally
>>
>>   todeliver: A-B-C-D
>>
>> He is happy with commits A, C and D but wants to reject B. Ideally I
>> want to be able to say
>>   git rebase --onto <parent of B> <child of B> todelvier
>
> You don't want <child of B> there, just B.
>
> git rebase --onto <onto> <upstream> <branch>
>
> Rebases the commits from the range <upstream>..<branch>, and that
> _excludes_ the commit (referenced by) <upstream>.
>
> So:
> git rebase --onto B^ B todeliver
>
> Works on: B..todeliver == todeliver --not B
> And that range contains commits C and D.
>
> Björn
>

Re: Removing a commit from a local branch

[Sitaram Chamarty]

On 2009-01-11, Chris Packham <judge.packham@gmail.com> wrote:
> Consider the following example. The maintainer has the
> following branch locally
>
>   todeliver: A-B-C-D
>
> He is happy with commits A, C and D but wants to reject B. Ideally I
> want to be able to say
>   git rebase --onto <parent of B> <child of B> todelvier
>
> and get
>   todeliver: A-C'-D'

you have an off-by-one error here.  You need B, not child of
B.  Problem solved :-)