On 2019-04-02 14:11:11 +1100 (+1100), Michael Still wrote:

wouldn't it be cool if we could somehow mark email from gerrit that came from a bot? Or inversely, call out email sent by a human somehow. That way I could null route all the bot spam and suddenly gerrit email would be useful again.

Is this a thing upstream gerrit supports at all?

The way CI systems and similar automation currently provide feedback on Gerrit changes is to authenticate with a regular account and submit a comment just like any normal user. To counter this, we long ago developed standard naming conventions for those accounts to simplify client-side filters. We have a couple of outliers ("Zuul" and "Elastic Recheck") but otherwise you should be able to trust that any account ending in " CI" or " Bot" is from an automated source. On my personal mailserver I use filter rules matching on "$USER has posted comments" in the message body (when found in a message with a "review.openstack.org" List-Id header) to catch the bulk of these. I gather Gerrit's upstream maintainers are working on a new mechanism specially designed for handling result reporting from CI systems, so this is probably a great time to engage and suggest that any E-mail notifications triggered by that feature include easily parsed headers identifying them as such. -- Jeremy Stanley