Gitlab experiment wrapup

[Alexandre Julliard]

Giovanni Mascellani <gmascellani at codeweavers.com> writes:

> Hi,
>
> Il 14/06/22 12:55, Alexandre Julliard ha scritto:
>> - Signoffs are a bit cumbersome (Rémi). We should change the
>>    requirements to something better adapted to Gitlab.
>
> I don't remember if this point has already been made, but we're using
> the Signed-off-by pseudo-header in a rather peculiar fashion. The 
> standard meaning of Signed-off-by is something like "I have checked
> that I have the legal right to submit this patch for inclusion" (more 
> specifically, see [1]), while we use it as "I have checked that this
> patch is technically sound". Other projects use other pseudo-headers
> for the latter, like "Reviewed-by" or "Acked-by".
>
>  [1] https://developercertificate.org/
>
> Could we take the occasion to switch to something more standard?

Signed-off-by is convenient because the git tools have direct support
for it, but the real motivation for us was to keep track of who
submitted the patch, to support sending a patch written by someone else.
In that case the convention with email is to set From: to the original
author, but the committer info is lost, hence the Signed-off-by.

With merge requests, the information of who submitted the merge request
is preserved independently of the commit contents, so the easiest option
is to simply get rid of signoffs completely.

Another option would be to require the submitter to explicitly approve
their own MR if it contains a patch written by someone else; that would
be similar to a reviewer pushing fixups and then approving the MR.  It
would require some mechanism to detect missing approvals though, so it
may not be worth the trouble.

Thoughts?

-- 
Alexandre Julliard
julliard at winehq.org