[1693] in SIPB_Linux_Development
Re: Redhat 4.2/Linux-Athena
daemon@ATHENA.MIT.EDU (Derek Atkins)
Wed Jun 4 14:03:11 1997
To: Erik Nygren <nygren@MIT.EDU>
Cc: Chris Murphy <chris@MIT.EDU>, linux-dev@MIT.EDU
From: Derek Atkins <warlord@MIT.EDU>
Date: 04 Jun 1997 14:02:43 -0400
In-Reply-To: Erik Nygren's message of Wed, 04 Jun 1997 13:06:02 EDT
Erik Nygren <nygren@MIT.EDU> writes:
>
>
> > We already discussed this on linux-dev, at length. I was fairly sure
> > we had consensus to move to an SRVD model.
>
> This was certainly not the impression that I had and was agreeing
> to. Although I'd like to see a SRVD, I think allowing people
> to install small portions of Linux-Athena is more important.
OH? Take a look at my message dated 27 May 1997 11:24:48 -0400 with a
subject of "Re: Making system packs an option in the RedHat 4.2
install procedure". Specifically, I state:
me> I personally prefer the "spm" model better. We can reduce the Athena
me> installation to two RPMS: AFS and Athena, where "Athena" contains
me> enough information to get spm up and running, and then it can run spm
me> to install everything else. Indeed, spm could probably even use
me> "synctree" if we wanted to be consistent with standard Athena.
me>
me> This would mean that there is a two-stage install: The Linux install
me> (which installs AFS and "Athena"), and then the Athena install, which
me> runs spm as part of the athena.init and installs the rest of the
me> packages.
Also, what percentage of Linux-Athena users actually care about
installing portions of Linux-Athena? On the other hand, how many
Linux-Athena users just use the small-gods installation procedure,
install everything, and are done with it? I believe that the majority
(and it's a LARGE majority) fall into the latter category, and only
the hacker-types fall into the former. I'm not sure exactly what your
priorities are, but I'd rather make the installation easier for the
majorit, even at the expense of making it a little more difficult for
the (small) minority.
> Examples of what I think is critical that people be able to do easily:
And how many people actually do do these things?
[snip]
There's nothing that says that we can't create RPMs... But that wont
be the main installation system, and I doubt that they will be kept up
to date.
> - I have a 2.1.x SMP or other WeirdShitHardware[tm] machine in
> my lab. I want to be able to run zephyr on it, but running
> AFS just isn't an option.
This is a problem. Then again, how many of these are there right now?
We don't support this hardware now, anyways, so I don't see how this
affects anything. See my last point.
> I strongly believe that packages should remain mostly as they are, but
> that we should find a way to make SRVD's an option. (Such as packages
This isn't what we agreed on linux-dev; go re-read my message and the
rest of the thread. The model is that everything is from the srvd and
users can choose to copy things locally off the srvd. Only two RPMs
remain: AFS and "Athena".
> full of symlinks.) One option would be to always make RPMs available,
> but to use the SPM system as the default way of installing things.
This is an option, but I see the RPMs falling into disrepair because
it's a lot of work to maintain them.
> But there really needs to be a way to easily install portions of the
> system *without* having to get AFS working (or even without being
> connected to the network). The ability to do this is actually one of
You have to be on the network to download the RPMS in the first place.
And Athena _IS_ a networked system -- you need a network in order to
use Athena. Therefore, I see no reason that "installation without a
network" should be a requirement of the Athena subsystem.
> the primary reasons I use Linux-Athena rather than NetBSD-Athena.
Ahh! Here is the reason for your complain. I don't think that our
energy is best spent trying to make life easier for the small minority
of Linux-Athena users at the expense of the majority of those users.
The majority are sitting in their dorms connected to MITnet and want
Linux-Athena on their PCs. They don't want to think about it. They
just want to see "xlogin" and get their zephyrs and email, and be able
to run Applix.
Most of these users don't care how the system works. They just want
Athena in their dorm room. Using the SRVD/SPM system makes this much
easier both on the developers (namely us) and on the installation
(which they have to perform). It also has the least number of
problems to provide multiple configurations of the system.
> Having a way to use a SRVD should be an option, but disk space is
> cheap enough and networks are slow enough that it shouldn't be either
> the default, or the mode that the system is designed around, IMHO.
Your opinion is so noted.
True, disk space isn't a major reason to use an SRVD. However,
multiply the disk space taken up by the srvd (and include TeX, since
that's going in there two) by the number of people running
Linux-Athena. That product is a large number.
However, the main reason to use the SRVD/SPM approach is simplicity
and upgradability. It makes it _SIMPLE_ for users to install. It
makes it simple for users to upgrade. This isn't possible with the
RPM system.
-derek
--
Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
Member, MIT Student Information Processing Board (SIPB)
URL: http://web.mit.edu/warlord/ PP-ASEL N1NWH
warlord@MIT.EDU PGP key available
