[90] in Release_7.7_team
Meeting minutes, 5/20/94
daemon@ATHENA.MIT.EDU (Dorothy Bowe)
Tue May 24 10:26:30 1994
To: release-77@MIT.EDU
Date: Tue, 24 May 94 10:26:06 EDT
From: Dorothy Bowe <dot@MIT.EDU>
If something seems inaccurate or down-right wrong, it's probably my
fault. Please let me know and I'll correct it.
dot
Meeting Minutes, Friday May 20
==============================
In attendance: Carla, JJ, Kevin, Gary, Mark, Dot
1. Build and integration report
a) RS/6000
We started off the meeting with a discussion of the RS/6000 build
situation. The latest and greatest compiler from IBM builds a
version of DES (and thus kerberos) which does not work. Since the
code has been around for a long time and been built on many different
platforms, we believe that it is a compiler bug. However, the compiler
used for last year's build existed only on a single machine and is no
longer available. Mark has therefore used Cygnus gcc to build a
working version of kerberos. However, the rest of the RS/6000
release will be built with the IBM compiler as we do not have time to
do otherwise.
The problem is that if we know the compiler breaks one piece of code,
we have to have some doubts about the rest of the release. In the
course of discussing this issue, we came up with several options:
1. Continue on to friendly test with a combination of gcc and IBM cc.
During friendly test we could attempt to get as much feedback as
possible on the IBM release to ensure that the system is being
adequately tested. At the end of the friendly test period we
would make a decision whether to continue to staff test, or hold
back the IBM part of the release for further testing.
[Note: There are approximately 10 RS/6000s signed up for
friendly test including a couple in the test cluster.]
2. Delay the IBM release and build the whole system with gcc. This
is not considered realistic for 7.7.
3. Repackage 7.6 with some selected subsystems (e.g. emacs 18.59, mh
6.8, perl 4.036) as 7.7, thus avoiding the compiler issue for this
year.
Since we don't lose anything by proceeding with friendly test, we
decided to go ahead and make a decision on a next step based on the
results from testing. (IMHO, at that point we could either go ahead
with the release or back out to 7.6+.)
Due to the complications from the RS/6000 build, the target date for
friendly test has slipped 2 days to May 31, 1994. At this point,
nothing else in the schedule has changed.
b) Kerberized Remote Access
Although Bruce was unable to attend the meeting, he contributed an
item to the agenda, namely an issue about the kerberized telnet and
telnetd which he has been working on. The particular issue which
began the conversation is that the presence kerberized telnetd will
not be invisible to users because you do not get tickets on the
remote system. Therefore, anything in your startup files which
requires authentication will not fail. The options are to educate
users, or to make a change to the global dotfiles which would detect
the lack of tickets and not source your dotfiles.
This led to a lengthly discussion of why this is so and what good is
a kerberized telnet anyway. If anyone is interested, contact me and
I'll explain. In the end, we reached agreement/understanding on
these questions. In particular, the point was made that this
release item is not a total solution to the general problem; rather
it is one piece in making future changes possible. (Such as
comp kerberized clients for the MAC and/or PCs.)
2. Friendly Test
The following (approximate) number of machines are in the friendly
test pool:
DECstations 16 (including 1 DEC 5000/133 and 1 DEC/3100)
RS/6000s 10
Suns 23
*** We need to have as many people as possible use the RS/6000s ***
An announcement will be sent next week to everyone signed up for
friendly test with instructions on how to update workstations. The
first update will be manual; subsequent ones automatic.
Action Items:
Dot will talk to Bruce or someone about how telnet[d] fits into other
projects.
Mark will send out email informing us if friendly test will be ready
for May 31.
Dot will prepare and send the message for friendly testers.
Next Meeting:
Friday, June 3 in E40-316