Marc Jones wrote:
If a patch is cherrypicked from gerrit and pushed again, it doesn't need a new signoff. I think that anyone that pushes a change should add a signoff.
Actually I think they need to - someone pushing the patch is supposed to say that they can, that's one reason for the signoff to start with.
It might help prevent the unintentional push.
Difficult. A hook that checks if the new commit is any different from the previous one would avoid duplicate pushes, but would also block updating commit messages, which could be desirable.
A few unintentional pushes do not matter IMO as long as each developer sorts out the configuration and procedures after discovering the mistake. Most are learning gerrit and more are learning git so of course there will be a few bumps. No big deal as long as we get them fixed I think.
//Peter
-
Peter Stuge