On Thu, 15 Nov 2001, der Mouse wrote:

# Um..._allowing_ characters that break NIS doesn't constitute breaking
# NIS, especially if we splash warnings all over the manpages - and
# perhaps at runtime; ISTM it would make sense to warn if NIS is enabled
# and such a username is added.

...true...

# But...what breaks?  I once had to work with NIS, and I don't recall
# anything in there that would break with nonprintables (except for
# newline and possibly tab...oh, and NUL).

Password changing broke for me.

# > I was able to change my password once to something with a control
# > char in it, but it wouldn't let me change it back because of that
# > control char.
#
# Odd.  Sure you were typing your lnext character where appropriate?

Didn't need it -- ^N doesn't usually need escaping.

# > I suspect one might run into similar problems with usernames.  If
# > someone can test this with usernames and prove me wrong, I'd love to
# > hear about it. 8-D
#
# I just now created a user called t\2u (C-style) or t\^Bu (vis(3)
# style).  Except that my shell won't let me use ~t^Bu to refer to its
# home directory (probably because it considers ^B a non-letter),
# everything seemed to work.  I set a password, logged in over a "telnet
# 127.0.0.1" connection, logged out....

Cool.  As long as there's warnings for NIS, it all looks good to me!

				--*greywolf;
--
NetBSD, Net Profit.