Going back to the __FreeBSD_version proposal, what I think would be
useful is first a full archive rebuild with this change, then see how
many packages break.
Then if they're only a few corner cases, enable it. But given how
slow are some maintainers to react and how long it takes for some
trivial fixes to be merged, I'd only do it at the beginning of release
cycle (you'll notice that Debian release engineering is a lot less
branch-based than FreeBSD's).