[1796] in NetBSD-Development
Re: Development on snork and cj
daemon@ATHENA.MIT.EDU (Marc Horowitz)
Thu Sep 24 21:50:06 1998
To: Derek Atkins <warlord@MIT.EDU>
Cc: Greg Hudson <ghudson@MIT.EDU>, Salvatore Valente <svalente@MIT.EDU>,
sipb-athena@MIT.EDU
From: Marc Horowitz <marc@cygnus.com>
Date: 24 Sep 1998 21:49:22 -0400
In-Reply-To: Derek Atkins's message of 24 Sep 1998 21:35:48 -0400
Derek Atkins <warlord@MIT.EDU> writes:
>> (Bcc'd to sipb-office, but moved to sipb-athena)
it was sent to sipb-athena@ihtfp.org, which doesn't exist. this
message is to: sipb-athena@mit.edu, bcc: sipb-office.
>> I think it's even worse than this, Greg. What we need is:
>>
>> * a stable machine per OS for 'current-release' development
>> * a quazi-stable machine per OS for 'new-release' development
>> * one shared crash-and-burn machine
>>
>> I agree that we don't really want to waste 5 head spaces for this.
>> However I do believe that we'll _want_ heads, at some point, for each
>> of these machines. For example, we want people to be able to use the
>> "current-release" machines, we want people to login and test the
>> "new-release" machines, and people need to see the crash-and-burn
>> machine to debug.
I think that you need two head spaces and a rack space. the c&b
machine can be in the rack. You can have one head space for each of
linux and netbsd, with a switcher between the stable and development
machines. Put the switcher on the development machine most of the
time (so it gets exercised) unless the release is really flaky, then
put the switcher on the stable machine. Of course, all will be on the
network all the time.
Marc
