[1570] in NetBSD-Development
Getting NetBSD-AFS back under control again..
daemon@ATHENA.MIT.EDU (Bill Sommerfeld)
Wed Feb 11 12:00:27 1998
Date: Wed, 11 Feb 1998 16:52:42 GMT
From: Bill Sommerfeld <sommerfeld@orchard.arlington.ma.us>
To: netbsd-afs@MIT.EDU
Cc: netbsd-dev@MIT.EDU
So, I now have a whole bunch of spare cycles and I can put them into
getting NetBSD on AFS into better shape.
I'm approaching this in full process-weenie mode just because I don't
like cutting corners... and because I want this to be as turn-key as
possible once I put an initial time investment into it..
Here's what I'm looking to do, not necessarily in exact order:
1) document the build process and release process, and keep
the document up to date with reality, so it can be handed over to some
other sucker by just saying "here, read this file and follow the
instructions"
2) get *everything* under CVS control. Sources, scripts, everything..
I want to get to the point where it's simply:
cvs checkout into empty directory
run script
wait N hours
and end up with an installable package at the end..
3) figure out what NetBSD architectures we really care about.
i386
sparc
m68: sun3, mac, ???
pmax
alpha?
4) figure out which NetBSD revisions we care about:
1.3
1.3.1
1.3-current
5) find build and test systems for all the relevant machines. My
understanding is that John Kohl had a few guest accounts for these.
I'm all set for i386 build & test, and have some hints on some of the
other platforms, but any leads would be very helpful. I'm going to
give preference to systems at MIT for this.
builder requirements: up & on the net 24x7, running at least 1.3 final
or newer.
tester requirements: running relevant release, with someone friendly
near the console to reboot it as necessary.
6) Do the builds.
7) Test all the bits.
8) Figure out the release process for getting the bits into the
Transarc cell for release to the outside world, and release them.
Once that's out of the way, we can start worrying about:
a) dealing with moving forward to a more recent AFS source base
b) looking at other goodies (kretch's encrypting AFS client patches;
UMich disconnected AFS, ...)
Feedback, comments, bricks, etc, welcome..
- Bill
