[prev in list] [next in list] [prev in thread] [next in thread] 

List:       git
Subject:    Re: [RFC] Proof of concept: Support multiple authors
From:       Christian Couder <christian.couder () gmail ! com>
Date:       2017-01-30 17:56:42
Message-ID: CAP8UFD3=vaFupEDay-5vrMBwK_YJezysUUvySxnUUZxuW7m_WQ () mail ! gmail ! com
[Download RAW message or body]

Hi,

On Sun, Jan 29, 2017 at 7:06 PM, Cornelius Schumacher
<schumacher@kde.org> wrote:
> This patch is a proof of concept implementation of support for
> multiple authors. It adds an optional `authors` header to commits
> which is set when there are authors configured in the git config.

I am just wondering if you have read and taken into account the
previous threads on this mailing list about the same subject, like for
example this one:

https://public-inbox.org/git/CAOvwQ4i_HL7XGnxZrVu3oSnsbnTyxbg8Vh6vzi4c1isSrrexYQ@mail.gmail.com/

> A new command `git-authors` is used to manage the authors settings.
> Authors are identified by initials and their names and emails are
> set in a `.git_authors_map` file.
>
> Signed-off-by: Cornelius Schumacher <schumacher@kde.org>
> ---
>
> When doing pair programming we have to work around the limitation that
> git can only have a single author in each commit. There are some tools
> which help with that such as [git-duet] [1], but there are still some
> limits, because the information about multiple authors is not reflected
> in the native git data model.
>
> Here is a proposal how to change that and implement native support for
> multiple authors in git. It comes with a patch as a proof of concept.
> The patch by no means is finished, it doesn't cover all cases and needs
> more tests and error handling. It's meant as an illustration of the
> concept.
>
> The basic idea is to introduce a new optional `authors` header in
> commits which contains a list of authors. The header is set in each new
> commit when there is an entry `authors.current` in the git config listing
> the current authors. When this config is not there the behavior falls
> back to the current standard behavior.
>
> When the header is there it is treated in the same way as the author
> header. It's preserved on merges and similar operations, is displayed in
> git show, and used to create a list of `From` addresses in `format-patch`.
> Email supports multiple `From` addresses as specified in section 3.6.2 of
> RFC 5322.
>
> When multiple authors are configured, they still write the standard author
> header to keep backwards compatibility. The first author is used as author
> and committer. In the future it might be good to implement something like
> automatic rotation of the order of authors to give credit in a fair way.
>
> To make it easier to work with the authors there is a new command
> `git-authors`. It sets the list of authors using initials as shortcut for
> the full configuration with name and email. The mapping of initials to
> names and email addresses is taken from a file `.git_authors_map` in the
> home directory of the users. This way it's possible to quickly set a list
> of authors by running a command such as `git authors ab cd`. This is
> useful when doing pair programming because the people working together
> usually switch quite frequently and using the command with the intials is
> quicker and less error-prone than editing the configuration with full
> names and emails.
>
> The command also supports setting a single author, setting more than two
> authors or clearing the configuration for multiple authors to go back to
> the standard behavior without the new authors header.
>
> The concept of the command and the mappings file is similar to what
> git-duet does, so that it should be familiar to many people doing pair
> programming. The behavior of git doesn't change when the new feature is
> not used and when it's used it should be backwards compatible so that it
> doesn't break existing functionality. This should make a smooth transition
> for users who choose to make use of it.
>
> Adding support for multiple authors would make the life of developers doing
> pair programming easier. It would be useful in itself, but it would also
> need support by other tools around git to use its full potential.

From what I recall from previous discussions, the most important
question is: are you sure that it doesn't break any other tool?

> This
> might take a while, but I think it's worth the effort.
>
> I'm willing to continue to work on this and create a patch which is suitable
> for inclusion in git.
[prev in list] [next in list] [prev in thread] [next in thread]