[1784] in Kerberos-V5-bugs
Re: SIGINT in the rlogind child
daemon@ATHENA.MIT.EDU (Sam Hartman)
Fri Jan 19 23:46:38 1996
To: donn@u.washington.edu (Donn Cave)
Cc: hartmans@MIT.EDU (Sam Hartman), krb5-bugs@MIT.EDU
From: hartmans@MIT.EDU (Sam Hartman)
Date: 19 Jan 1996 23:46:20 -0500
In-Reply-To: donn@u.washington.edu's message of Fri, 19 Jan 1996 16:56:07 -0800
>>>>> "Donn" == Donn Cave <donn@u.washington.edu> writes:
Donn> By the way, I made the following change to login.c, for
Donn> Ultrix 4.3. It was using the wrong terminal discipline,
Donn> which also had the effect of killing the session on <ctrl>C,
Donn> though that's probably just a coincidence.
In the future, please send bugs and patches to
krb5-bugs@mit.edu. That section of code looks fairly broken and has
not been changed, so I'm forwarding your message to someone who
understands old BSD terminal handling better than I. I am concerned
that while your patch looks reasonable for Ultrix, it is important
that the set line discipline ioctl not be handled at all on Posix
systems.
Donn> This was version 5 beta 5, with authtf patches for DCE.
I would like to clarify that MIT has only released one patch
since beta5 and that patch dealt with a security problem in telnetd.
There are no other official Kerberos5 patches; several unofficial DCE
patches have been made, but you should probably be more clear about
where you got non-MIT patches in bug reports, so we can determine
exactly what code you are running.
Thanks, though for bringing our attension to a rather broken
section of code; it appears it needs some touching up to meet our
standards regarding using autoconf tests whenever possible. We try to
avoid defines based on particular vendors' system whenever possible.
--Sam
