Re https://pantsbuild.slack.com/archives/C046T6T9U/p1645750979476569?thread_ts=1645750208.194249&cid=C046T6T9U If there's valid reason to violate the official deprecation policy, and it seems there is here, then imo that's reason to revise the policy a bit rather than just violate it seemingly arbitrarily. That the current policy may have led to persisting a "super confusing" situation for a while suggests to me that something is worthy of a bit of revision here. Thoughts...?

In this case though couldn't the env vars be deprecated and during the deprecation cycle just be merged with the blessed env var subsystem, potentially erroring if there is a conflict on merge to keep the deprecation promise; i.e.: not trample the deprecated env vars?

Yeah, it is certainly theoretically possible to get the deprecation working. I couldn't figure it out though and I exceeded my timebox. Maybe I could get it with fresh eyes