[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Resignation as seul-install leader
SEUL is not going anywhere so I am resigning as leader of dev-install.
Why SEUL is a failure.
Ten months after the beginning of this project we are about at the
same point: vaporware. An unqualified failure specially compared to
the Stampede project: they needed one month work to put together a
distribution. Why of this situation?
I see several reasons: a wrong delopment model, dreams, politics,
The development model.
SEUL was started with an announcemnt. Tens of people came to the
list. Each one posted his little idea. Other people answered. Then
a newcomer arrives and posts the same idea who have been already
discarded after being discussed to death. And discussion begins anew.
Succesful projects don't start with an announce: something is done by
a small, closed group of people, sometimes by a single man. When
there is something working then and _only_ then an announce is made.
This way architectural decisions don't take forever thanks to the
limited number of people and the fact than there are no newcomers
restarting old discussions.
The something working must be small, exciting and usable.
Small, because at this time the developing group is not big enough
for an extended project and because the longer the preliminary phase
the greater the danger of the developping team losing heart and
Exciting because in phase 2 it is important to have more people
working on the project for expanding it: the program must be able to
raise enthousiasm. At this point the working something imposes a
frame so there will no be endless arguing on architectural issues. In
addition the originators of the project have gained a _natural_
authority based on work done so when a discussion is going on forever
they are able to settle the matter knowing than both sides will
willingly accept the decision. Linus leadership in kernel matters
comes from the fact than he is the author and from his amazing
achievement. Of course fascist leaders, people enjoying power or
quite simply dumb leaders are a bad thing: people flee the project so
no progress is done. Nor when nobody can take decisions.
Why the project must have a working something? Because the people
with a genuine interest in improving the software are the people using
it. If it doesn't work, no users so no developpers, so no progress.
The Wine project is an example of this: it was already work in
progress in 1994.
Long term wild dreams are a bane on a free project: they make lose
time on discussions and in addition are very bad for moral when people
begin to realize how far is the goal. In this list there were
assertions about how SEUL would make Microsoft go bankruptcy thanks to
the wonderful user interface Linux would have thanks to SEUL. Small
problem: not a single line of anything had been written and poster had
not even defined what would be that interface. There were even people
discussing about how to share the money coming from SEUL sales and SEUL
I remember you how Linux started: Linus was making some expereinces on
task swtching so he wrote two small tasks one writing 'A' and the
other writing 'B' to the display. He found it easy so he started
writing a small OS for the fun. It was 32 bits because this was the
hot feature in 386. First Linux version did not have virtual memory.
Then when more and more was being done Linus became more ambitious for
his system and added VM, when people came to Linux then X, TCP/IP and
many more were being added. Linus did not start yelling: "I will
write a system able to reach world domination." He would have been
sent to the nearest hospital.
The lesson: start with small dreams and make them bigger when you make
progress. By the way despite being alone Linus wrote a kernel with
virtual memory and ported several key programs including BASH (a tough
nut) all in nine months. About the same time SEUL has been around.
That is what you are able to do when you don't lose time with
discussions and wild dreams.
That has been a plague in this project from the start. We have seen
countless people hyping about Debian but never caring explaining us
why they were thinking Debian was good for an end user. I am also fed
up about the harassment of other distributions: civil wars in the
Linux community only weaken it.
What I find absolutely shocking is than too many people in this list
are acting like Trojan horses for Debian instead of thinking in the
good of this project. From the start they tryed to have SEUL based on
Debian without questionnning if that was good for it. And now they
want to make SEUL a Debian satellite just like former communist states
respective to Soviet Union. Remember that post a few days ago about
SEUL being more adequately named Simple End User Debian. Linux
disappearing of the name: symptomatic isn't it? Instead of working
for Linux expansion SEUL according to the poster would be a mere
intoduction to Debian. That would solve a major problem for that
distribution: because Linux is a minority OS, beginners are the key of
expansion and Debian attracts few Linux beginners. But this project
was created with a greater goal than serving Debian
The problem is than for Linux expansion a project like SEUL is TEN
THOUSAND times more important than Debian. A primary factor limiting
Linux growth is than Linux can only attract users able to survive in
it. Lowering the expertise level needed for using Linux and trying to
make it about as easy as Windows is fundamental for Linux future
because it increases its potential user base. That means than TRYING
TO USE THIS PROJECT AS A TOOL FOR PROMOTING A DISTRIBUTION IS HARMING
We didn't come to this project in order to help a distribution gaining
domination in Linux world. We came because we wanted to help _Linux_
reaching world domination.
Irrational decisions: the Debian backwards leap.
It seemed logical to take the distribution the nearest to our goal and
just work on it. For copyright reasons we cannot use distribs like
Caldera, Suse or Craftworks because some essential parts in them are
proprietary (we still can take the packages they make of free software
however). That leaves us the three GPLed distributions: Slackware,
Debian and RedHat. Slackware with its lack of packaging system and
minimal administration tools can be discarded. About Debian if you
want to know the spirit of the distrib just look at the many
installation floppies, at the six web servers and no Wysywyg word
processor, at the way the kernel is compiled (minimal features
despiyte being bigger than the one in most other distributions: little
care in avoiding recompiling for the users) and you will find than
Debian has been designed with programmers in mind. Not end users.
"To know the taste of the sea you only need a mouthful" (Solsjetnitsyn
speaking about Gulag). I drank from Debian several times. I know
about Debian integration. This is useful in an end users distribution
but integration is also a programmers goal: #include <stdio.h> applied
RedHat like Caldera or Suse have been designed with companies in mind.
Not end users. But in a company having a person spending a day
RTFMing or customizing his working environment costs around 150$ a
day. Either you make a Linux who can be used out of the box (no
kernel recompilings for instance) and relatively easy to configure or
Linux becomes quickly _much_ more expensive than NT. Even if RedHat
sucks in many ways for a home user it happens than easiness to
configure and usability out of the box are needed both for companies
and for end users. Unlike Caldera or Suse, RedHat GPLs its work so we
can use it. About the many bugs in .0 Redhats (.1s are a lot
cleaner), one month after RH release you ftp to your favourite site,
then go to the RH update directory, mget *, disconnect, "rpm -U
*". And you get a distribution as good as any and miles ahead of
Debian or Slackware in easiness of use.
All that translates in than for making an end user distribution there
is less work to do starting from RedHat than from Debian. Omega agreed
about that but unfortunately he had been lured into dreams of Debian
help and 200 programmers coming to our help like a legion of
archangels. All is at stake on these 200 archangels. What happens if
they don't come? What happens if they are of little help? We would
have compromised the SEUL project by making it start from afar and
forcing us to reinvent many wheels. We would have compromised the
most important project for Linux future.
Will the 200 archangels come? Bruce Perens was very evasive "we will
be looking and if you are worthy we will come, perhaps we will do the
job alone". In addition in a mail with him when I expressed concern
about the lack of end user software in Debian he answered me "People
package the software they find interesting, that is why we have six
web servers but no Andrew or Thot (free Wysywyg WPs)." (*) That means
than there are good chances than, even if Bruce summons them, only a
small minority of the Debian team finds interesting to work for SEUL.
I also have strong doubts in the real value for us of 200 people who
work for four years and don't find time or interest in including that
fundamental piece of sowtware for an end user: the most mature of the
free Wysywyg word processors (Andrew). By the way Andrew is now
proposed to Debian developpers willing to package it. I don't know if
my talks with Bruce Perens are the cause. However a week ago Andrew
was still an orphan.
(*) Bruce authorized me to quote him despite my warnings than it could
be used against Debian. My sincerest admiration to him.
About the Debian way and the usefulness of SEUL.
Imagine than thanks to the 200 Debian archangels SEUL is released.
And then what? Our target users are _not_ going to download a
distribution from the Net. They will need a CDROM. Is it easy to
find Debian on CDROMS? You can get it from Cheapbytes and LSL but
Infomagic who included Debian in its 97 editions is not including it
in the February 1998 edition. And Joe User is not going to buy from
Cheapbytes: the only way to make him use Linux is putting a big nice
box in the stores than Joe User uses to buy from. He will see
RedHat, Slackware and Caldera. NOT Debian.
Our Joe User has got his SEUL CDROM and then what? Putting a CDROM
without paper doc in the hands of Joe User has 99% chances of having
him failing and going back to DOS. A wise beginner does not buy a
1.95$ CD from Cheapbytes for his first contact to Linux: he either
buys a boxed Linux with paper doc or a book about Linux describing a
distribution included in the CD accompanying the book, possibly both a
book and a boxed Linux. In fact he will probably go Linux because he
heard about it and stumbled upon a Linux book or a Linux boxed
edition. I only know of one book talking about Debian. It is a thin
book published by an obscure editor and difficult to find. In
contrast nearly every computer book publisher has at least one Linux
book in its catalog, and you can find them in about every book store
with a computer book department. And what is included in those books?
Redhat, slackware, sometimes Caldera. NOT Debian.
Every computer geek reads English. Joe Users don't (unless than
they live in an english speaking country of course). A translated
Linux is fundamental. Slackware has been translated. RedHat has been
translated. Not Debian.
I wanted to show the huge obstacles to overcome until we put a
Debian based SEUL in the hands of end users, and that means than
because unexperienced users are getting Redhat not Debian we would
spend years before being of any use for them. And being of any use
for Linux. Now remind than beginners are using Redhat and suppose
SEUL was compatible with RH: an RH (or Caldera) user who finds hard to
use the software provided by RH could just be pointed to the SEUL site
and download the package he needs. RH provides sendmail, our user
could get an easier MTA from us, RH does not provide a small DBMS or a
Wysywyg word processor we could provide him with Andrew and Xmgrok.
And after SEUL being of help for many users some would try the full
thing and then word of mouth would _perhaps_ have a book publisher
willing to put SEUL in the CD of one of his books. But we would be
_immediately_ useful to the Linux cause instead of having to wait for
years before we are able to put SEUL in every book store. Too bad for
those Linux beginners than sendmail has led to the verge of returning
Microsoft. No help from us: we are going Debian and there are no
beginners using Debian. Too bad for Linux.
SEUL is a structure too heavy for useful work at this stage. In
addition idelogy and politics not reason are dictation decisions.
When I tried to reason my fellow leaders I was unable to have a
serious discussion about the SEUL future. A rational discussion was
impossible: I received a hail of hype and mystique about Debian and
nothing more. At one point I was so exasperated I proposed each of us
criticizing his distribution just to insure than leaders were lucid in
what had to be done and were able to look objectively at their
distribution. Besides me only one accepted the challenge. That
speaks a lot.
I resign from dev-install. I submitted to my partner the guidelines
for an installation. He will publish them. I hope they will be
useful for this project.
Linux is the only thing that matters and SEUL is more important than
Debian for Linux future.
Jean Francois Martinez
The worthy man is the one who would drink muddy water if such is the
water of truth.