[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
SEUL: Of methods and goals and mice and men
FIRST PART: How to make a free software project a failure or a success.
There is one common point in successful free software project: they
had few developpers at first but they they started witth modest goals.
But that allowed them to have something working quite fast. So they
got users and because users had the source code some of them became
developpers and then with their help the project expanded and achieved
far more ambitious goals than could have been dreamed. Linus used
that method, and the GIMP team. HURD did not use it: not usable, so
no users, so no developpers, so still not usable.
So first release of SEUL must have modest goals but achieve them.
SEUL's goal is not fixing UNIX world but improving significantly upon
RedHat or Debian from the easiness of use point of view. We must hack
an existing distribution instead of starting from zero.
We must keep software developping to a minimum. Existing
distributions can be easily improved with a combination of removing
some key road blocks and better software selection. Let do that now
and in the second edition we could be more ambitious.
SECOND PART: Operational goals
Some people will disagree with the examples I give, please do not
flame me about them. Flame me if you disagree with the goal they
served to illustrate.
1) Start from the easiest to use distribution.
The closest the distrib is to our goals the lesser the work and the
farther we will able to go.
With this criteria we should start from LST but this distribution is
harder/more expensive to get than RedHat or Debian, so it will be
difficult to put it in the hands of developpers.
2) Improve installation (can anybody be against this :-).
Instead of asking people about their LAN card and their IP address (so
assumming you have a permanent connection to the NET) installations
should ask about where is the modem or what is the phone number of
It could be a good idea to allow people to try LINUX without
repartitionning. That means UMSDOS for the root file system, and
loopback files for the remaining (UMSDOS is too slow and space hungry
for general use). Of course allow them also a a classic installation.
3) No problem is unsolvable if you can get help.
That means than a major goal for SEUL is allow people to get mail
and news out of the box, and that for a realistic setting: assume the
user has no permenent connection to the INTERNET. Also if possible
provide connectivity for users of private networks like COMPUSERVE or
AOL. If you can't post and get news then you are unable to ask for
help about how to fix this problem. Unless you prefer having them
reboot W95. :-)
4) It is not enough to make LINUX easy to use, we must gave people a
good reason to use it.
We must put in it software than you cannot get in MS world, at least
not at affordable prices.
GIMP comes immediately to the mind. I also want a WYSYWIG Word
processor (THOT if it is stable), a modeller (AC3D), a spreadsheet
(Wingz). Yes AC3D and WINGZ are shareware in LINUX world but I think
it is valid to include sharewares or commercial software you can use
gratis for a limited time (long limit) if it is good, its
policies are acceptable and we can't find a good free substitute.
5) Do not allow tradition to interfere.
If we feel than something traditionally used in UNIX world
interferes with our goal then we must find it a substitute.
In every distribution I know, when an app needs an editor VI is
started. In SEUL the EDITOR and VISUAL variables must be preset to
point at an easier one. And I don't give a dime for the "VI is
present in every UNIX", we will put it (to keep UNIX integrists quiet)
but it must not be the default editor: we must avoid new users being
confronted with VI unless they invoke it explicitly.
6) No cod liver oil drinking.
There are some ordeals in UNIX/LINUX you are expected to take but
are no longer necessary.
One instance is the traditional kernel recompiling when you finish
installation. You are not considered a real Linuser until you have
done it. In fact a good distribution can make this basically useless.
With good use of modules we can avoid it: this is no longer 1.2.13.
If you really insist in making them compile the kernel give them a
kernel compiling proc where the user has only half a dozen questions
to answer: the non-modularizable things like processor type or PCI
7) A good user interface.
KDE is by far the best available but it will not be ready in time.
The goal is than we must provide the beginner with the means to avoid
the shell if he wants. Tools like file managers are available but we
must put them under their noses.
8) Config tools.
Two parts in this: configuring the box and configuring application
software. We must try to get what configuring tools are available in
LINUX world. I think it is a pity than most distribs do not include
Dotfiles: it supports a limited number of software and the UI is not
so great but it is better than to configure by hand.
Also when choosing a WM I find highly desirable it has scripts for
building automatically its menus when starting so user will not see
uninstalled apps or miss the app he installed the day before.
Also the user must not be forced to RTFM if it has normal needs. It
must work out of the box, get installer software or some prebuilt
configs one one of them covering his needs.
Natural reaction when you see an ugly program with Xaw and default
(black on white) settings is to think than if the programmer was
unable to make an effort on visual quality then probably he was lazy
when programming. And that reaction is more acute between people with
little computing experience.
Out of the box apps must come with resource settings to make them look
good. We should use Xaw3d as soon it has catched up with Xaw (it felt
behind when Xaw upgraded to 6.3), or still better neXtXaw. Also
whenever possible replace Xaw apps by equivalents using Qt.
We need a manual. The venerable Installation and Getting Started
needs a face lift, we could give it this lift and ensure than it
centers on SEUL.
Also people in this project look at Woven Goods it is simply
We also need to ensure a way than if an application installs HTML or
info docs they are automatically linked into the general tree.
11) Putting things under users noses.
He must get a file manager (tkdesk?) when he starts X. User must not
be forced to discover by himself than he has one. We must ensure than
he will not do things by hand because he does not know we gave him a
user friendly tool for doing it. That means than this apps must be in
X menus or automatically started apps.
A demo would be useful to show him what is available in the
distribution. There are people who reboot DOS just because they
simply don't know they have a LINUX tool sitiing in their disk and
able to handle the task. Because you get a lot of things in a Linux
CD it is easier to forget them than paid-for Windows apps.
Jean Francois Martinez
==================== The Linux. Use the Linux, Luke! =======================
Simple End User Linux Mailing list
To be removed from this mailing list send a message to firstname.lastname@example.org
with the line
in the body of the letter.