Re: SEUL: Re: SEUL distribution?

[jfm2@club-internet.fr]

   Reply-to: omega@seul.org  <<<-- !!!

   On 16 Jan 1998 jfm2@club-internet.fr wrote:

   > Precisely.  I have been on this list from the foundation of SEUL and I
   > never saw a such thing than a Debian review.  In fact _I_ was the one
   > who published the reviews about Redhat, POwer Linux (alias Caldera
   > Lite) and I think also Linux Universe.
   Yeah, I thought you did those.  I guess I was wrong about the Debian
   review.  Well, I'll post and see if I can get anyone to do one.


   > Only answer we got from the Debian partisans was hype, not reviews.
   That has been a problem, but then that's been the problem with RedHat as
   well.  This is partially why I want to build a core distrib and be done
   with it.

   > -How Debian installs?  Dselect  :-(.
   That will be changing.  SEUL is in the right position, with e-Linux as
   well, to help Debian's Deity project become the best installer Linux has
   ever seen, and easily outdistance anything M$ can come up with.  The key
   is to work together and make sure the project is generic enough to work
   for all distribs, but simple and stable enough for actual users.

   > -How Debain configures?  That is not bad with the concept of
   > configured and unconfigured packages.  :-)
   I like that idea, though I don't know if a SEUL machine would be able to
   utilize it very much.  Depends.

Very useful: you know what has not been configured and you can ask to
be propmted for all of it.

   > -How good is the DEbian user interface?  You don't even get
   > auto-upgradable menus.  And you get mosly raw X apps so the user is
   > supposed to twinkle with resources.  :-(
   And RedHat is much better?  Yes, they have put some work into developing a
   few window-manager setups, and some menu automation, but have they tried
   to build a set of tools and applications to flesh out the interface?  Not
   that I've seen.

A real GUI for Linux will not be possible until GNOME.  About RH
against Debian: having menu entries point at empty air confuses end
users.  And getting raw X apps is also bad: an end user does not know
enough or have time enough (real work to do) to hack resources: he
must get something acceptable out of the box.

   > -Is kernel compiling mandatory?  (A no-no for SEUL).  Yes (little use
   >  of modules).  :-(
   As Bruce mentioned, the 2.1 series is almost completely modularized.  For
   those things that can't be modularized, such as what processor it's built
   for, there's a really simple fix: give them a 386sx kernel (what RedHat
   gives you right now), and a Pentium-optimized kernel.  Easy as that.

   Of course, I'm not even remotely advocating using the 2.1 kernel for SEUL.
   However, I think 2.2 will be out and reasonable stable before SEUL is
   ready to be released, so we can plan for it.

There are imperfections in 2.0 modularization but problem is not in
2.0 versus 2.2.  The real problem is in what is the attitude about
kernel compilings.  If the authors of the distribution say: "Kernel
compiling is not a problem for real men" then you end with a big
kernel (not than it is a big problem on a 16M box to spill 500K) and
so slow at booting due to hardware probing than the user will be
compelled to recompile it as soon as he gets a login prompt.

If you think than a beginner has too much things to learn at the same
time and than kernel compiling is dangerous for him then you go as
modular as possible.  If you ship only one kernel you cannot get an
optimal kernel but in 2.0 you can get one good enough than the user
can live with it.

So we have than with all distributions shipping 2.0 all kernels are
not created equal.  Kernel compiling is a very good test for knowing
the spirit of the distribution.

About this theme look also at the propositions I made for a special
cut-down but simpler kernel compiling procedure.

   > -What software is included?  In theory there is much in Debian.  In
   > practice I found than there is much software for computer phreaks (six
   > Web servers, oh god don't they know the law of decreasing marginal
   > utility?)  but is below average in software more or less suitable for
   > an end user.  :-(
   This is what a core will be used for.  SEUL will not be distributing all
   of Debian, plus some of our own work.  It will be distributing the core,
   plus the relevant layers, which will be common to Debian, SEUL, e-Linux,
   etc., plus our own work (i.e. applications, etc.).

   > My overall apreciation on Debian 1.3 is than as a starting point for an
   > end user distribution only Slackware is worst.
   It depends on how much work is being put into it vs. how much you want to 
   get out of it. I agree that RedHat may be the logical choice to start from
   if we wanted to get something out the door in 6mo.  However, RedHat has

Six months from now using RH according to your schedule, longer still
using Debian?  Noooo!  Understand than time is a factor playing
against us.  Understand than the first non computer-oriented people
are coming to Linux: look at some questions in the newsgroups to be
convinced.  Understand than each time a newbie gets burned into one of
the existing distributions he becomes a person difficult to persuade
trying anew (burn me once, not two) and one who will tell horror
stories to his friends.  Understand than we must keep Linux momentum.
Understand also than with UNIX in the feeble hands of SCO, the
disappearing of the X consortium, with NT having the advantage than
now standard (not high end) PCs can run it, NT has a good chance of
becoming the standard OS for PC in place of W95 and using this base
for crushing commercial Unixes in a few years.  It is essential than
by this time Linux be strong enough (with many _end_ users) it is able
to survive stand-alone.  Now it isn't.  SEUL should be shipping _now_
not in six months and having SEUL delayed until 1999 would be still
worse.

Also remember than gothic cathedrals needed centuries to be built.

And than Hurd was already late before Linus got his 386.


   its problems, and given that RedHat is an actual company, working with
   SEUL to better their own distribution at the expense of their own plans,
   well, that's just not gonna happen.

See the RHAD project.

   With Debian, we have the opportunity to work together with the largest
   distribution team around, and they are heading in some of the same
   directions we are.  Why do it all alone if we can get another 200ppl to
   help us out???

Depends on what means working with Debian.  If that means beacoming a
Debian satellite with Debian following its own path and SEUL trying to
cope then benefit is nil.  If SEUL exerts an influence of Debian and
the Debian people start thinking in end users even without centering
on them like SEUL then that would be very positive.  Not only for
SEUL.  The "top-seller" Linux distributions (now Slack, RedHat and
Debian) are responsible of the image of Linux.  If SEUL influence
makes Debian easier and Debian pay more attention to end users then
the whole opinion of the public about Linux will improve.

An illustration: Debian comes with six web servers.  I am not sure is
that useful but that is Debian problem.  For SEUL it only confuses the
user having so many.  But despite its 200 ppl Debian 1.3 came AFAIK
without any WYSYWYG WP.  It would be good for Debian, for SEUL and for
the whole Linux cause if instead of adding a seventh web server the
Debian people concentrated in adding Andrew or Thot to their
distribution (they are both free).


I have a last observation: an important factor of succes is
associating people to the project and the way you treat them.  Every
one having received an announce for something as important as a change
of base distribution?  I haven't and it seemed I was not the only
surprised.  So if this is not a mail lost problem I find such things
highly unfortunate.

-- 
			Jean Francois Martinez

==================== The Linux.  Use the Linux, Luke! =======================