[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SEUL: Observatons

> On 24 Jun 1997 jghasler@win.bright.net wrote:
> > winston writes:
> > > Who is SEUL directed at?
> > 
> > I think it should be directed at the individual user.
> I'm going to take one more stab at convincing people that the first step
> needs to be geared towards a LAN/Workgroup/Intranet environment.  If most
> everybody on the list continues to summarily dismiss this point of view, I
> guess I'll drop it.  No use beating a dead horse.

And I'll take one more stab at trying to refute you ;-)

> 1) I don't think people reinstall operating systems lightly.  Those who do
> can probably handle RedHat.  


Well - isn't the whole point that RedHat *is* too complicated from
an end-user standpoint?  In terms of sophistication, Windows NT is
on par with a full-blown linux system, but an end-user buys it, 
runs the install, gets the option to configure a few devices, 
lets it do its 15 or so reboots, and then reaches for the Quicken
CD to install that.  When they look at their disk, they see:
or something like that.  When this same user is done installing
linux, they have no notion of C:, they have no idea what /usr
/home etc etc is all about, and all the files in /etc are completely
mindbogglingly complicated.

OTOH, a workgroup in a heterogeneous environment looking for a 
server, hears about RedHat for $39 (vs. NT server $xxx), and see
a real benefit to taking the time to install and learn it.

The idea of SEUL is, in my opinion, to bring out a distribution
that *introduces* linux.  Someone gets a copy, or downloads it,
starts installing it, and are met with a positive experience.
If the user likes it, they can go to a lot more sophistication
by installing RedHat or something like that.

The bundle of software that came with OS/2 was very well received
because it gave the users the feeling they could do something.
At the same time, this is not the end-all, since we have to 
try to mainly attract users with a certain level of curiosity,
otherwise we give ourselves a support nightmare.

So - my $.02:

1. A nice install.
	I personally think we should assume that a FAT partition
exists on the PC.  We should provide a boot manager that runs from 
this partition (the partition could be tiny) - but something more sophisticated
than lilo.  However, we should install linux into its own partition
(no fudging with running from a FAT partition).

I know some people are going to hate this, but I think the install
should be in two stages:

1. boot from floppy.  The kernel supports EIDE (also CD) and
has modules for most SCSI controllers.  I strongly suggest we aim the 
first install disks at the most generic setups - ie. IDE disk/CD
and/or SCSI disk CD.  Lets leave out all the different MFM/RLL/Mitsumi
CDROMS etc.  Do autodetection for EIDE/CD first, then SCSI, if the 
kernel does not autdetect SCSI, give the user the option to select
SCSI controller from a list and specify parameters.  Then let the user
set up partitions, and format FS.  *Create a swap file* don't make
them create a seperate swap partition - seems like a waste to many 
novices.  Then copy vanilla install system from CD/FAT partition/floppy
onto the partition. Reboot.

2. Boot manager config in FAT partition (self-configuring
utility). Reboot

3. Boot into linux setup.  Starts stripped-down vanilla VGA
Xserver by default for config tools which will allow setup of X,
PPP, users etc. etc.  Tell the user about the X three-finger salute
and try the X configuration.  If it fails, have a way to revert to 
the generic VGA server again.  When it is done, present the user
with the "tour of linux" by HTML.  (I just know the X-thing is 
going to give me a lot of grief - but at the same time, I don't
know anyone who doesn't run X)

2. Management - we might as well hack away at RedHat's control-panel
stuff.  It works pretty good.  One thing it's missing is that you should
be able to fire it up as a normal user, and then have it prompt for a 
password to let you do the management.

	I also think we should have a database of available programs
that the user can peruse, and tell them to go visit sunsite or redhat
or debian (or the CD)for that matter to get packages.  This implies that the 
installed system uses RPM (or somesuch) but I disagree.  I think we
should build it from packages and ship it as a finished setup including
the package databases etc.  (ie. you don't need rpm to install the 
base system)

3. We might as well throw in nedit (statically linked) as the editor
and someone else can come up with the word processor we should get.
Some will say LyX, but I'm just going to come right out and say that 
I think LyX is a totally misguided effort. 

Simple End User Linux Mailing list
To be removed from this mailing list send a message to majordomo@txcc.net
with the line
unsubscribe seul-project
in the body of the letter.