[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
The Think Linux HOWTO
The following is one of the more important links of the projected new
site. It details the basic differences betweeen the Linux and Unix
user. IMHO the main problem with present distributions is that their
designers have been educated much like traditional Unixers and lack
basic understanding of the problems of the Linux user. Think in all
the times you have read "Contact your system administrator": the
writer obviously didn't even think it was possible to be a system
administrator from firrst second.
=========================== BEGIN ======================================
<HTML>
<HEAD>
<TITLE>Think Linux</TITLE>
</HEAD>
<BODY>
<h1>Thinking Linux</h1>
<p>There are many ways to consider Linux. You could look at its API,
at its basic utilities and consider it is just an Unix clone. But you
could look at the fact Linux is free and runs on unexpensive hardware
and then you will see an operating system who has the potential to
become a system for personal computers and for dektop computers
provided it is adapted to a different environment and to different
users.
<H2>Linux for personal computers</h2>
<p>
It could seem obvious but the user of a personal computer is not at
work doing tasks mandated by his boss. His motivations are having
fun, hobbies, managing his household. When thinking in such a user
games you have to think in games, software for image or sound
creation, for house automation or personal finances. Everything who
can be fun or useful for real life. Web server or C compilers are
unimportant here, at least for "normal" users.
</p>
<p>
Users of a personal computer don't have an experienced system
administrator caring for them until they become grown up users. The
simplest tasks can be difficult for them, RTFM is not an answer for
their problems: there is a critical phase when the user will be unable
to find the doc, don't master the tool for reading it and will perhaps
be unable to understand the docs. Here we need a robust system who
forgives user's mistakes. We need to learn to put essential info
under the user's noise (something Unix programmers regularly forget)
provide sensible defaults and tools for simplifying system
administration even if those tools don't scale well to large sites.
</p>
<p>
Many users of personal computers will never see another Unix than
Linux in their whole life. They have no special reason to use such or
such program just because this is the "traditional Unix tool" so
forget about traditions and about your personal tastes and provide
what is best for the user.
</p>
<p>
The strong "hackeristic" background of Linux should not make us forget
about people who have real work to do and whose goal in life is NOT
learning computer science. Linux has office suites as good as the
ones available for other systems so people willing, say, write a
thesis could do it on Linux but if we don't learn to provide
distributions who are as ready to use as possible then they will use
Macs and Windows for this kind of work. It is specially important to
get rid of the "kernel recompiling myth". Since kernel 2.0 it is
possible to provide canned kernels supporting all but the most exotic
Linux features while the performance difference between a canned
kernel and one compiled by the user is negligible. So there is no
excuse when a distribution provides a kernel who, due to lack of
features or poor performance, will force the user to recompile it. It
is this kind of things who makes people choose other system for real
work.
</p>
<P>Private users connect to the net through dial-up links. It is
vital to allow PPP configuration as soon as possible. We also have to
provide for mail and news configuration and ensure traffic is routed
when the link is brought up. <em>For most home users the PPP link is
the only way to ask for help</em>. You also have to think that in
many countries phone is expensive so proxies should be used to make
connections as short as possible.
</p>
<H2>Linux in small organizations</H2>
<p>Small organizations deploy a small number of computers and while
inbig ornizations training costs could be recouped by license savings
on a large number of boxest hey will not in small organization. So we
have to ask ourselves if some traditional servers like sendmail or INN
are not overkills in such orgnizations. And provide a range of
"typical" prebuilt configs for Samba or netatalk.
</p>
<H2>Linux on the desktop</H2>
<p>At times it looks like if Linux designers were unconsciosly
persuaded that no sane person would use Linux on the desktop. Unix
lost the desktop war not beacause it was user hostile (it lost the
battle in DOS times and DOS was as user-hostile as Unix) but because
it was expensive. This made that no sane businessman would use Unix
for taks, like WYSYWYG word processing, who could done on PCs. And so
those companies who tried to port file managers or word processors to
Unix found that sales were disapointing.
</p>
<p>
But Linux is cheaper than Windows or NT, it is more stable and in
addition in our times of internetted computers the lack of security in
Microsoft software is becoming a serious problem. There is no reason
it could not invade the desktop if we learn to think in productivity
software, in office suites and get rid of some Unix features like
manual mountings who were designed for servers not for workstations.
</p>
<H2> Conclusion</H2>
<p>
The present Unix-like Linux is perfect for a fraction of the Linux
users, but we should not allow being coerced into keeping Linux in the
narrow Unix boundaries. There are other users -less influential and
less vocal but potentially biggr in number- whose needs are different.
Linux growth needs we adapt to those users and we can do it while
providing traditional Unix-like interface to traditional users. We
will have to learn to think in <em>all</em>Linux users. We will have
to learn to <em>Think Linux</em> and keep our minds
<em>independent</em> of traditions
</p>
</BODY>
</HTML>
======================================= END ========================
--
Jean Francois Martinez
Project Independence: Linux for the Masses
http://www.independence.seul.org