Re: stgit: managing signed-off-by lines

From: Karl Hasselström <kha@treskal.com>
Date: Jan 25, 2008, 14:30 UTC
Message-ID: <20080125143026.GA10132@diana.vm.bytemark.co.uk>
In-Reply-To: <9e4733910801250534s7ab11b15uff3c782165a8e2ea@mail.gmail.com>

On 2008-01-25 08:34:53 -0500, Jon Smirl wrote:

Show 13 quoted lines

> On 1/25/08, Karl Hasselström <kha@treskal.com> wrote:
>
> > What command are you using to export patches?
>
> I am using "stg export" from one branch and then "stg import -s" on
> another. After I import the patch and do 'stg edit' on it, it has
> the signed-off-by line in it. When it is export from that branch the
> signed-off-by get added again.
>
> You don't want to strip the signed-off-by lines on import, they may
> be from other people. I would think that on export you could look
> and see if my signed-off-by line is already on the patch before
> adding it again.

I haven't set up my templates to add sign-offs on export; instead, I give --sign to "stg import". This always does the right thing:

  1. It's smart enough to notice if my sign-off is already there, and
     don't add another copy.

  2. My sign-off is propagated even if the patches are transferred by
     pulling instead of by email or "stg export".

For patches that I write myself (so that they are never "stg import"ed), I add the sign-off manually in the commit message, with the aid of this elisp snippet:

  (defun sign-off ()
    "Insert sign-off line."
    (interactive)
    (git-append-sign-off (git-get-committer-name) (git-get-committer-email))))

Adding the sign-off conditionally in "stg mail" and "stg export", as you suggest, could of course be done. But that would have to be command line options, since the template system can't really express that kind of transformation.

-- 
Karl Hasselström, kha@treskal.com
      www.treskal.com/kalle