Keith Moore wrote:
The argument from somebody saying that his "by name"
API hides IPv4 vs. IPv6 details as it should,
The presumption that the API "should" hide the
differences between IPv4 and IPv6 is a dubious one.
IPv4 and IPv6 really do act differently, for a
variety of reasons. Some applications can get away
with pretending that they're the same, others cannot.
Okay, oversimplification on my side, the argument that
impressed me didn't include an "as it should" part, it
was about using the same old code with a "by name" API,
Bill mentioned that IPv6 should have got its own new
class, something I cannot judge, certainly no job for
For a stupid direct-to-MX script I've sometimes used
it just happens that its "by name" API is limited to
IPv4, in theory the code would use "implicit MX" also
for IPv6. Willing to fix this when 2821bis says so:
IETF mailing list