Other than the "rules are rules" point of view (which does have some
point), I don't see why not to add it, either now or soon. The only
practical drawback is that there might be some bug with this new
feature that we haven't discovered yet and will, in theory, make the
5.0.0 release look bad if some new developer finds it. But, really,
that's not very significant, is it?
And surely since it can't possibly break anything, this handy feature
should at least go in for 5.0.1? And if 5.0.1 is ok why not for 5.0.0
right now, unless perhaps you want to wait for 5.0.1 to see that it
gets extra testing? It doesn't need to wait for 5.1, however many
unknown months that is away, does it?
And no I don't think it's particularly reasonable to expect users to
be back-porting stuff from Head to 5.0 all the time. If you're going
to recommend that as being easy (as opposed to "if you really need it
you can try, but I think it'll be tough, that's why we haven't done it
yet"), then why not just "back port" it sooner for everybody to use,
like right now? If it's not good enough to go into the stable branch
than presumably it's not good enough to recommend for people to
back-port either, no?