[SCM] Samba Shared Repository - branch master updated

[Martin Schwenke]

On Tue, 24 Jan 2023 17:54:30 +0100, Andreas Schneider via
samba-technical <samba-technical at lists.samba.org> wrote:

 
> Oh and I have a review check in for my 'autobuild' command.
> 
> https://git.cryptomilk.org/users/asn/dotfiles.git/tree/git/.gitconfig#n113
> 
> line 113-119
> 
> This will warn you if there are unreviewed patches.

I think this will error with no possibility of override.

I agree that a review check in a hook is a good idea.

However, before doing that, we need to clarify our rule:

  Each commit in master should have been reviewed by at least two samba
  team members. For this purpose, involvement in patch creation is
  considered as reviewing.

  - https://wiki.samba.org/index.php/CodeReview#policy

There is also this:

  Outside contributors are welcome to review patches also, this is a
  good way to learn more about Samba! 

  - https://wiki.samba.org/index.php/Contribute#Review_process

To me, a sane understanding of is that each patch must contain 2
Signed-off-by: and/or Reviewed-by: tags by different Samba team members.
Checking for 1 Reviewed-by: tag isn't sufficient (for non-team
submissions and if there are non-team reviews) and we will occasionally
still mess up.

What we really need is a script that has access to a list of Samba Team
email addresses (which are already public knowledge), and counts the
number of Sign-off-by: and Reviewed-by: tags containing those addresses.

However, there is still a corner case.  :-(  Notice I said "different"
above?  I have many WIP commits in branches that are years old.  If I
update and contribute some of those now, they will have my personal
Signed-off-by:, to acknowledge the circumstances under which they were
initially written, and a DDN Signed-off-by:, to acknowledge current
circumstances. In this case, we could de-dupe using the name, so the
process would become:

1. Get Signed-off-by: and Reviewed-by: lines
2. Keep the name for those with a Samba team email address
3. Count the number of unique names
4. Is it 2 or more?

Any other corner cases?  :-)

peace & happiness,
martin