summary (to those who does not read source-changes):
	- finger does not let 8bit characters through.  it convers anything
	  with 2^7 bit set into M-x on sending (fingerd), and strips off
	  2^7 bit on receiving (finger)

	- kim is trying to make finger 8bit thru.  kim changed sender/receiver
	  side check to isprint(), which is locale dependent.  kim says LANG
	  (or LC_CTYPE) setting should be sufficient to change behavior.
	- itojun is arguing that: (1) protocol-wise it has to be ASCII by
	  default, from RFC (2) we need an explicit agreement between sender
	  and receiver to use non-ASCII charset, so there has to be a
	  protocol extension (define "charset" directive and submit i-d),
	  (3) the change has to be backed out until #2 is accomplished,
	  (4) isprint() change does not consider multibyte case at all.

>But the fact is that many finger daemons send out 8bit characters, and

>reading Finnish with the 8th bit stripped is really difficult.  Other
>finger implementations allow the 8bit characters, and I still don't see
>the danger that you seem to find in the current implementation.

	as far as I checked FreeBSD and OpenBSD has the same behavior
	as NetBSD (of yesterday) - escape any non-ASCII chars with M-x on
	sending, and strip off 2^7 bit on receiving.

itojun