code-review
code-review copied to clipboard
Decide if we want to fallback on "in-patch" instead of latest revision by date
With latest revision by date, in case the parent revision of a patch is not public or was not analyzed and stored in the backend for some reason, we can overpublish.
E.g. this happened in https://phabricator-dev.allizom.org/D3533, where the revision couldn't be found on the backend (since the revision is from staging m-c). In ab25e7dfc0dc5bb8217fb436aac0957faf03a793, the issue was there. In ffee8e3a7870875ab78af162638548f72cdabddf, the issue was not there anymore because it was fixed. Andi's revision is based on ab25e7dfc0dc5bb8217fb436aac0957faf03a793, so the issue is still there in his revision. When we get the latest revision by date, we get ffee8e3a7870875ab78af162638548f72cdabddf (or newer), and so the issue is not there and so we consider the issue as new.