In the "Scaling PostgreSQL Development" unconference session. One of the problems that came up was that people don't follow "best practices". The response to that was that people don't know what the best practices are (nor that they are important to follow), because we don't enforce them. Based on the discussion there I'm planning to make the cfbot fail to apply a patch in the following two cases:
1. If a patch is not created by "git format-patch" (but cfbot will still use "patch" to apply the patch in case "git am" fails)
2. If the commit message has no body (so only a title)
Does anyone have strong opposition to this? To be clear, it means we don't run CI on patches created by piping "git diff" to a file anymore, as a way to nudge submitters into providing useful commit messages.
Communication wise, I plan to show this in the CF app as "Fails apply" (instead of "Needs Rebase"). When clicking the "Fails apply" link, it would then show a log as to why the apply failed. need to be created using git format patch, and should have a descriptive commit message.
