[3156] in testers
Patch release 8.2.2 out
daemon@ATHENA.MIT.EDU (Greg Hudson)
Tue May 27 02:03:57 1997
Date: Tue, 27 May 1997 02:03:23 -0400
From: Greg Hudson <ghudson@MIT.EDU>
To: rel-eng@MIT.EDU
Cc: testers@MIT.EDU
Athena patch release 8.1.2 just went out for both platforms. Here are
the changes:
* (Solaris) A bug has been fixed in dm where it would corrupt
the passwd file when a user logged out.
* (Solaris) /usr/lib/fs/fd has been pulled local to make
/dev/fd work.
* (Solaris) Root's path has been modified to avoid letting
commands like "shutdown" come from /usr/ucb.
* (Solaris) Formatted man pages have been installed in the os
volume for /usr/share/man, /usr/openwin/man, and
/usr/dt/man.
* (SGI) The update script for 8.1.0 has been modified to force
SGIs to rebuild their kernels, which is important for the
new AFS to work. (This change was also applied to Solaris,
but it doesn't change the behavior of the script for
Solaris.)
* (Both) TeX knows how to find itself now.
* (Both) lert works now; a bug has been fixed where it tried
to use getservbyname() to locate a server.
* (Both) bogusmsg.ps has been reinstated as part of the
Transcript installation.
* (Both) The named.root configuration file for the local name
server has been updated to reflect a change in location of
one of the root name servers.
* (Both) A small clarification has been made in the
documentation in named.boot on how to make local records.
* (Both) The update script renames rc.conf to rc.conf.orig
when it makes a change, instead of rc.conf.old (which is
used by mkserv).
* (Both) The post-reboot phase of the update now runs
config_afs so that it honors setuid bits on the packs.
There are still some unfixed bugs:
* There is a bug in attach such that NFS filsys records
receive higher priority than they should.
* 8.1 Suns spit out a bunch of named error messages at boot
time; they should be suppressed.
* The krb5 rlogind KRBTKFILE default is different from
rkinitd's; possibly they ought to be the same.
* Using "more" at the reboot prompt fails irrecoverably, which
is bad. Something should be done about that if possible.
* An SGI, wellfleet, has been known to panic in conjunction
with xscreensaver state changes.
