[Précédent (date)] [Suivant (date)] [Précédent (sujet)] [Suivant (sujet)] [Index par date] [Index par sujet]

Energies need to be focused



Something disturbing happened at Montreal's Installfest last October. Though
Linux made lots of progress for the last two years, the attendance was hardy
one tenth that of 1999. Has Linux been hyped too soon? Is there a problem
with the development of Linux for the general public?

Everybody knows the motto "Release Early, Release Often". But how early is
early, and what should be released? It seems many developers wouldn't be
caught dead releasing a soft with less features that comparable Windows
applications. But Windows applications have been adding features for years
and even the first... well, at least, the more serious ones  -- Eudora, Free
Agent, Arachnophilia, etc. --, came out with not so many bugs from the
start.

When the newbie is confronted to 10 bugs in the first 5 minutes, he's rather
annoyed. Whereas a programmer working on the kernel would try applying a
fix, the newbie wonders how come the developer hasn't found the bugs in 5
minutes too. He then tries another browser, email or news client, etc. and
gets fed up. So, why don't the distros first choose the applications that
work?

Of course, this appears as an intrusion in the "normal" course of Linux
events, but wouldn't it be an incentive for developers to present a cleaner
product? If ever a distro made a biased choice -- choosing Webmin instead of
Linuxconf, or vice-versa, for example -- another distro would offer the best
tool and take the advantage. You know how things go: people talk on the net.
After a while, instead of having 50 email clients, we would most probably
end up with energies concentrating on 5, and Linux would end up MUCH
stronger.

And even with applications that work well, what sense does it make to ask a
newbie to choose between vi, vim, pico, joe, elvis or... emacs?  What good
is it for him to install them all? Wouldn't it be more logical to offer ONE
text editor, give the basic instructions -- for instance "You enter the
editor in command mode. Type help, press Enter key, you'll get all you
need -- and let him do the choosing while he gets accustomed to Linux?

Of course, all other applications that are judged valuable by the distro
designers could be included on the CD for LATER installation. But the fact
is the newbie wouldn't be left making guesses for two hours while
installing.

Then, let's face it, the distro designers don't seem to be very much into
beta testing. The expert -- he runs a database server for his store -- who
installed Slackware 8 on my computer spent 4 hours figuring out why the
&&%$#@ modem didn't work. Had he read "Running Linux" as I later did, he
would have found that /dev/modem had to be pointed to another COM port with
the ln command. (This being said, IMHO, Running Linux is certainly not a
book to help beginners to run Linux. I just lucky on this one.)

Two weeks later, I installed a new monitor with xf86config and the modem!!!
stopped working again. I could see no cause - effect relation here and
couldn't figure out what the hell was going on until I explained my mishaps
to a friend. When I came to the first problem with my modem, I figured that
xf86config repointed /dev/modem to the wrong COM port and finally found out
that AUTO didn't work for the serial ports in the BIOS. To settle the
problem once and for all, I had serial port 1, where my mouse is connected,
point to COM1. Since I had nothing on serial port 2, I set it on DISABLED.

And that was it! With all those distros claiming to make the beginner's life
easier, how many explain serial ports can't be left on AUTO? Not one, I'm
afraid, unless they managed to hide the info deep down in their site.

Who makes tests to find out what kind of problems the newbies are confronted
to? I mean, I'm not asking for any DrakConfs here, even fdisk is just fine
for me. Cfdisk is already a luxury I'd rather do without. Parted also does
the job. No DrakXXX needed! All I need is clear instructions.

I then tried to read a floppy in KDE. I read the instruction coming with KDE
2.1.2 . It didn't work. I went to KDE site, made a search, found something
in the FAQs... completely different from the instructions in KDE itself(1).
In a certain directory certain files could be found. Well... they didn't
have the same extension, but one did have a floppy as a logo. However, the
tabs didn't match that of the instructions. I tried matching the tabs best
as I could, it didn't work.

(1) But there was still no mention that you had to create a directory to
mount the floppy. Apparently, you had to know that beforehand.

I also had to get my fax and CD-RW working... For the CD-RW, I found long
instructions at Mandrake's, but I figured it wasn't even worth trying since
it seems the Read-The-Fucking-Manual thundering experts can't write. I had
already had lost too much time in this rat maze and went back to Windows.
It's sickening, but I can read a diskette from a GUI.

Of course, I've been told I lack determination. What is it to spend one or
two days linking a floppy when it is to run Linux for ever and ever, amen?
I'd wholeheartedly bow to this affirmation, if only one Linuxie admitted
that in this decadent world of ours, 99% of the population lacks
determination even more than I. In the present conditions -- remember this
last Installfest in Montreal? -- "ever" could come to an abrupt end pretty
soon. Before transferring all my data to a new OS -- address books,
mailboxes, favorites... batch files!!! -- I'll wait until there's more than
a handful of cheerful young devotees switching to Linux in the city.


Recently, Redmond Linux -- "the 45,000 square foot company" -- came out with
a selected applications distro at about the same price as Mandrake's basic
distro. RL pretends it has worked two years on its distro, to check that the
mouse is on the right com port, that the floppy is readable, etc... At 1,000
square foot per employee -- which means a 50' x 20' room for each! -- that's
45 employees! Hey, Patrick The Man Volkerding, what's your opinion? Aren't
they a little bit overstaffed?

Still, their proposal is dead on: a simple Linux distro that works right out
of the box. It's been my proposal for more than two years now, while I was
being told "RTFM!' But rather than going for "the 45,000 square foot
company", I'd prefer one with more modest means : ) Slackware, for instance.

Slackware has made a move in the right direction with 8.0 -- pkgtool might
not be apt-get, still, it gets you there -- but why not streamline and take
more attention to detail, you know, just having a real dummy like me install
the distro before it is sent on the market?

Am I ever dreaming of this little jewel case with two CD and a little
booklet for installation, like Turbolinux used to put out! Cheap to produce,
hardly 4$, and cheap to send.

Once the desktop is installed, more on ipchains, a switch from "su" to
"sudo", installation of a database or an html server, etc., all on the CD,
with the search command helping for reference. One thing at a time, no
rhetorics, just practice, kinda,  "Now, here's how to chmod your directories
and files on your site." There's nothing like learning when you have a goal.
Then, for searching on peculiar topics, you do put an index at the end. And,
once again, you choose the applications that work well.

I believe this will be the future of Linux distros... or should be, if Linux
is to live. As people will get faster internet access and distros are
available for a little download fee, it will become apparent how silly it
was to put up companies to distribute Linux and to offer centralized
services as in a classical capitalist scheme. That's not the idea behind
Linux.

The idea behind Linux is to share knowledge and decentralize, not to provide
fish, but to teach how to fish. Then those who have the interest to go
beyond the desktop would provide services to companies and get paid, just as
any worker or professional. For now, those people are rare and costly
because the base is not widely educated. After the best of the crop has been
picked, the industry's hunger for competence remains unsatisfied. That's why
companies rely on Microsoft and it's silly rote learning OS.

For now, two distros have been wise enough to keep away from financial
turmoil: Slackware and Debian. Though they do not have the largest user
base, I believe, they constitute Linux's future. Maybe, as a Quebecer, my
gallic sentiment makes me lean more towards Slackware: I just luv how The
Man and his little team manage to resist the invaders. But if Debian ever
manages to put out a distro for general use with clear instructions less
than 18 months after a kernel is out... who knows? I know Ben Collins is
doing his best to accelerate the process and both distros appear as a valid
choice.

Recently, one of those sapheads who flourish on newsgroups in pretty much
akin versions pretended there was "no itch", Linux was doing fine in
Montreal. 10 persons at the last Installfest, Linux-Expo turned into a few
seminars for tax breaks at l'École Polytechnique, a booth and conference
room offered for free at the last Comdex left empty, no LUG in Montreal, and
we were doing just fine! And it seems pretty much everybody prefered this
rosy version of reality, because there was but one counter-reply.

When will this I-couldn't-care-less attitude stop? Progeny and Stormix are
defunct. Corel-Linux has been bought but is still in limbo. TurboLinux is
doing... well, certainly great business down under against Red Star. SuSE
has recessed to Germany. Mandrake, left without a franc after profligate
spendings, is now openly collecting money on its site. Caldera is
contemplating migrating to a BSD licence. VALinux has sunk to unfathomable
depths. Only Red Hat is faring well after being the first to lure
investors... that is if their suits for illegal manoeuvres on the
introduction don't succeed.

Of course, I am told, those earthy matters don't concern Linux, which is
pure spirit in essence. But what if it wasn't? What if its performance could
be evaluated by the almighty dollar? Would it cater to anybody but, mainly,
ISPs? And, after the trials are over(1), with Microsoft's user base ever
expanding, uncle Bill's satellites and uncle Paul's high bandwidth network
distributing their picture databases and rent-it-all services including
..NET, doesn't anybody expect any
"I-need-an-NT-server-for-my-Frontpage-extensions" kind of argument creating
a tidal wave? What about a Passport argument, then?

(1) Since the 11th of Sept., I believe the priorities have been made clear:
don't touch the shaky economy. So the issue is ineluctable.

Wouldn't it be good for the Linux community to think marketing in the
noblest sense of the term? Is Linux gaining market share? Does it cater? Are
the efforts invested bearing fruit? If not, and even if not a cent was
involved, just in sheer terms of lost time, Linux would turn out to be the
largest bankruptcy ever.

The stakes are also the highest ever. In a world whose data exchange would
be controlled by a monopoly and so-called national "governments", democracy
would be just an empty dream. Getting rid of this Brave New World
dictatorship would be next to impossible. It would have to die from its own
contradictions and the resulting soft carnage -- you know, people dying on
the streets -- would make the last world wars appear as jolly picnic
parties.

I'm not a programmer. I'm a writer... turned activist, I guess. But if this
community is minimally open-minded, surely, it will appreciate my opinion,
as it has always been poles apart from Eric S. Raymond's, the renowned
"evangelist". I never thought Linux would have 750 millions users in a snap.
I've always been looking forward to a fierce struggle, for which we'd better
get organized, as Microsoft will never accept anybody to stand in its way.

If ever an organization was in dire straits, Linux is. I'm completely
befuddled by some experts kindly losing their time advising newbies for
years on how to replace lilo in the MBR and then pretending they don't have
time to put up a clean desktop distro that would get their community
involved in the process. The mere experience would teach them a lot in terms
of teaching, writing, administration, marketing, politics. (There's so much
more to Linux than programming!) Darned xmodmaps for legacy and enhanced
French-Canadian keyboards could be installed! In the long run, more
contributions would be made to Slack and Debian. The Canadian and Quebec
governments could be sensitized to the issue of open standards. It happened
in France, why not in Quebec... or the US?

Touting anarchy as a development model is out of season. Forget the Bazzar,
anarchy never got Linux anywhere. Nothing is more organized than a bunch of
programmers working on a project on the net: it's the very best way to
separate the wheat from the chaff. When a "product" gets on a market, where
there are other contenders, another methodology is needed. For Linux, the
formation of companies who took responsibility away from the base, has been
the worst setback imaginable. With most companies nearing exhaustion, the
community now looks like a bunch of boy-scouts fighting with slingshots
against B-52s.

Energies need to be focused. And the first goal is the desktop and the
community.

Gilles Pelletier