[RFC] Developing a policy for diagnostic updates

[AaronBallman]

@cjdb

Informal seems fine to me. Getting a bit of consensus

Fantastic, SGTM!

@cjdb

So the reason I made this post is because I’d been struggling to understand what a handful of recent diagnostic reviews were trying to achieve, and D153359 happened to communicate the information in a way that very clearly helped me understand it. Most of these changes involved diagnostics that span a few lines, so it was probably difficult for me to hold all the pieces in my head at the same time.

I’m less interested in it being in the commit message, and more interested in it being associated with the commit (i.e. not just a comment in Phab). Release notes meet this criterion as well.

Excellent, so it sounds like we basically should be reminding code reviewers that release notes should be expressive and show diagnostic changes when possible?

Btw, to be clear, it’s always reasonable when reviewing code to ask for examples of what the ramifications of the patch are. Feel free to ask for more details any time you feel they would help you provide a higher quality review.