Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 1 Apr 2021 09:43:52 -0600
From:      Warner Losh <imp@bsdimp.com>
To:        Matthias Fechner <idefix@fechner.net>
Cc:        freebsd-git <freebsd-git@freebsd.org>
Subject:   Re: Ports Repocopies
Message-ID:  <CANCZdfoUAP1g9U=1Yzp4Yk9v%2Bsj6ACsiyZcNZwpHyL6CEmoA1Q@mail.gmail.com>
In-Reply-To: <e3f23b18-ba92-fa50-b130-3cae9b81349a@fechner.net>
References:  <202103311842.12VIgiJ3006095@slippy.cwsent.com> <20210331184820.GA26869@freefall.freebsd.org> <202103311859.12VIxlcn006294@slippy.cwsent.com> <6cf11fc8-d75e-451d-ada2-03a9f06048b4@www.fastmail.com> <e3f23b18-ba92-fa50-b130-3cae9b81349a@fechner.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Apr 1, 2021 at 5:07 AM Matthias Fechner <idefix@fechner.net> wrote:

> Am 31.03.2021 um 23:54 schrieb Dan Langille:
> > I have read a discussion of repo-copy, perhaps on the git mailing list.
>
> if you copy a file/directory and diretly stage it after the copy, git
> should automatically detect that it was copied.
>

If you do this, then Git is more likely to get it right, but Git is
fallible.  One feature of the ports tree is that it has lots of files which
are only a little different from each other.

However, there's two things that mitigates this design decision. First,
files it gets wrong have so little data to preserve that when Git gets
things wrong, recovering from it is usually trivial. Second, we're planning
on adding extra data to the commit messages to thread the history back when
git can't do it automatically should that be necessary after Git guesses
wrong and we need to get to the history. It's believed this will be fairly
rare.

Warner



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfoUAP1g9U=1Yzp4Yk9v%2Bsj6ACsiyZcNZwpHyL6CEmoA1Q>