gerrit changes: Forge author behaviour - coreboot

24 Feb 2014


      Hi,
I extended our gerrit installation to support "forge author" safely.
Forge author is the capability to push changes with an author field
different from your own. This is useful when sending patches you've got
from someone else while preserving attribution.
One downside of this capability, due to which we disabled it for a
while, is that it's possible for a change to go through gerrit without
the original author noticing. Maybe the author had issues with the
change because it was still unfinished - but good enough for others.
This is what happened for us.
So we now have a new column in gerrit, "AMR" for "Author-Must-Review"
that's active if a change is sent with different author and committer
fields.
Under the new policy, the original author must review the change
positively (meaning +1 or +2) before submission is allowed. Of course,
the usual rules (Verified +1, Code-Review +2) still apply.
There is one notable exception: Since there are some large upstreaming
efforts every now and then (mostly by Google), this rule is inconvenient
and counter-productive there. We can assume that within such groups,
people allow their "designated upstreamer" to push changes for them.
Because of this, the feature allows to add teams, where people can
publish changes for each others without the new requirement.
If you have such needs, please tell me about your team and I'll set it
up.
This is in testing mode - if you find any weird behaviour or you run
into walls because of this change, please tell me.
If that model doesn't work out, we'll revert it. But for now, these are
the rules.
Regards,
Patrick