We and our users might be happier if the (surprisingly new) version_cmp function, from GT#7504 were backported to the stable series.
TJ, please review for suitability (safety) for backporting. I expect the merge to be clean; if is, feel free to push it after approving.
I propose we cherry-pick backport only the two commits that add functions to fncall.cpp, but not the changes that avoid having a second copy of natural compare. Also the one block of text documenting the new functions should be hand-backported to the stable manual. Those three things should backport cleanly.
Cherry-picked those two-commits. Merged cleanly.
Cherry-picked the block of text’s commit because it was easy to throw away the changes to the version history file that doesn’t exist in the stable branch.
Implemented :jira: in Sphinx, added release note using it.