>Surely we could at least limit the damage to people who want to build
>software written to 2292 to something like -DIPV6_RFC2292 (which would
>then make everything switch back to 2292's API)?
>
>That's sure a lot better (for almost all values of "better") than
>telling people with N-zillion-line third-party programs "sorry, get the
>author to rewrite it all to this new RFC that came out last month".
>(Okay, maybe not last month by the time this happens, but you get the
>idea.)

	in RFC2292 backward compatibility, there are stuffs we can never
	support (for instance, type 0 routing header loose/strict bit,
	which was dropped in RFC2460).  i really would like to avoid "#ifdef
	WANT_2292" in headers.  the # of customer of RFC2292/3542 API is very
	small.

itojun