Re: SEUL: RFC: Base SEUL on Debian?

[Rick Jones <rickya@siservices.net>]

Erik Walthinsen wrote:
> 
> After talking with Bruce Perens quite a bit about this subject, it sounds
> like our best bet will be to start developing from the existing Debian tree.

I'm curious why you aren't just starting from scratch?  I am a Debian
user, have been since .9x version, but I still don't get the issue
concerning which dis' to use as a foundation.

My original experience with Linux was in pulling it off sunsite, I
believe it was, from scratch.  It was a pain in the ass since I was a
Linux dummy at the time but I was able to get things up and running,
including X.  I decided to go with a dis' because every time I tried to
compile something it would blow-up on me since it needed something else
not explained in the README for it, or it wanted to find things in a
different path.  I figured a dis' would solve those problems.  Not!  

My mistake was getting Slackware, 1.2 I think it was.  This was not much
easier than what I had before.  I went to 2.2.0.1 which was marginally
better, then I found Debian.

My point here is, to my knowledge, the reason there were so many
problems using source intended for say slackware with say Red Hat, was
the path's were not the same.  Other than being ELF or a.out.  So now
that the file system is fairly standardized what, other than reading the
package file formats, is the problem here concerning compatability?
 
> As soon as SEUL starts to pile stuff on top of this core, I would like to see
> a branch split off of the core to start incorporating all the Debian futures,
> including COAS, Diety, and possibly a few *basic* pieces of GNOME.

I heard that Linuxconf and COAS are talking of combining somehow.  You
might want to check into that.

> Eventually, I would like to see this core distribution become standard,
> possibly with Debian itself using it and helping to maintain it.  If
> acceptance grows, I would like to see other distributions using it.

I know you have been talking with Bruce about this, I don't know who
else, but Bruce is no longer in charge of the Debian project.  I don't
see why Debian, or any other dis', would use SEUL as a core since it
would no longer be a distribution in it's self.

> It is my belief that Linux can only go so far with a fragmented set of
> distributions, each of which is somewhat incompatible.  Vendors must
> currently compile and test their apps on Debian, RedHat, Slackware, Caldera,
> etc., if they want it to be used under Linux.  This is a major reason why
> companies don't want to invest in a port, as simple as it could be.

Again.  I have remained Debian for a long time and haven't bothered with
raw Linux or other dis's since, but unless there has been some drastic
changes someplace the only problems going from one dis' to another
should be the libc version, a.out or ELF, kernel version, and filesystem
tree, other than Motif or something non-linux like that.  This tells me
there needs to be a standard in versioning and filesystem trees, not a
new Linux dis' that everyone should follow.  Like they would.
 
Red Hat has already shown you that they aren't going to play ball.  I'm
not against your new dis.  I think it's a fantastic idea to make Linux
dis' that even a CI Windoze user can work with.  But I think you need to
be realistic in your goals here.  Commercial distributors are in it for
the money.  They could care less if there's a unified Linux.  They get
the software for nothing and sell it.  They make money, bottom line.  If
all the dis's get inline as you are proposing, their profits are going
to drop, if not disapear all together.

The only way a commercial dis' is going to follow your lead is if they
become the odd man out that can't run commercial software packages.

So your goal, using Debian as a base or starting from scratch, should be
to make SEUL so user friendly and popular that the commercial dis's jump
in line to remain popular themselves.  So don't follow, lead!

This is by no means intended as a flame.  I just sub'd lastnight and am
having trouble following your goals since, as far as I can see, the only
way your going to make a universal Linux is to include an installer that
can work with the other dis's package formats.  Other than that it
doesn't matter what you do, Linux is Linux.

If I'm wrong, please, enlighten me.  I'd like to get involved in a
project that wants to make Linux the choice of the CI user and blow M$
away.