| Weekly edition | Kernel | Security | Distributions | Search |
| Archives | Calendar | Subscribe | Write for LWN | LWN.net FAQ |
Show me the code!
Show me the code!
Posted Sep 15, 2005 7:42 UTC (Thu) by hingo (subscriber, #14792)Parent article: UserLinux: Autopsy
Great article! This is a topic I'm especially interested in: why os projects succeed or why they fail. UserLinux in particular has been interesting to follow.
Personally, I've always thought that UserLinux' fate could be seen already in the way the project was started. Bruce simply did it the wrong way. (Sorry Bruce, who would have known...) If we compare with Ubuntu and virtually all other distributions that exist today, they always code first, then start with announcing something that exists, that then will further be be improved.
What happens when you do thing's the UserLinux way is, you get a lot of people to join your mailing list. These people are interested in discussing what an ideal distribution should look like or what it's name should be. The people interested in trying out your distribution (also known as developers and users!) have nothing to get from you. In shortyou attract the wrong kind of people. It was clear that this group would not succeed in delivering a serious distribution.
Same is true for the name thing. Once you pick a name, be prepared that it will stick no matter how bad it is. Without a BDFL, no amount of mailing list discussions will ever get the name changed.
For the same reasons I disagree with the KDE analysis above. You brought this problem on yourself. You invited people to debate the issue of which DE is the better one! And even worse, to pick the only one to be used. For me personally, as a KDE user but not developer, this was the junction after which I knew for sure, that even if UserLinux might have ever delivered something, I would never use it. This is what the people on the mailing list have tried to explain to you, even though I wasn't one of them. (I just stuck with my current distro that provides a good KDE desktop, which again is a better thing to do than to whine on a mailing list.) That's what you wanted, that's what you got. Again, compare to Canonical. They made a decision only to provide Gnome, even before anyone knew about Ubuntu. When Ubuntu came out, that's what it had. That posed a challenge to KDE fans. They knew no amount of crying on a mailing list would change things, but they could accept the challenge and produce an Ubuntu with KDE -> Kubuntu. Now people can try both and actually decide for themselves whether one is better than the other. Again, if you have something to say, show me the code.
In short, UserLinux was an experiment in design by committee. I don't blame Bruce for anything though. I agree with the view he then held, that he is the person who could do this kind of thing. Now that he couldn't do it, we just know that the concept really doesn't work. (Which we kind of knew already.)
(Log in to post comments)
Show me the code!
Posted Sep 15, 2005 8:33 UTC (Thu) by ryanthiessen (guest, #29436) [Link]
In addition to what you say, even as a Gnome fan I could see that Bruce tilted the "debate" from the very start to produce a Gnome selection. I think you are correct, if only one was to be chosen it would have been healthier had the choosing been overtly done very the very start. Then it would have attracted and detracted from the start instead of alienating much of the interested parties.
Show me the code! (UserLinux)
Posted Sep 15, 2005 21:41 UTC (Thu) by Duncan (guest, #6647) [Link]
Good analysis!Something always bothered me about UserLinux, but I could never really put
my finger on it, tho you'll see me trying as far back as the comments to
the original LWN announcements on it. Finally, you put your finger on it.
My complaints centered around the fact that it seemed to me like it was
developed mainly over a case of bad feelings over RH's moves, and I
thought that was the wrong reason/way to go about founding a distribution,
but I couldn't point out /why/ it seemed that way to me. Your analysis
supplies the missing causative: at the root, I was uncomfortable with all
the pronouncements, without even the beginnings of the code behind them to
lend them support. The result was that of the pointing finger effect -- a
message (unintended and entirely innocent I'm sure) of willingness to
condemn someone else without being willing to "walk the talk".
So.. I'm glad you were able to put your finger on what I could never quite
figure out...
Meanwhile, the practical effect on the project was much as you outlined.
A project being born without code and without even many of the underlying
assumptions worked out was an invitation for all sorts of idealists and
do-gooders to sign up, while both by the same token and from the result of
the first, proving rather repellent to the practical sorts of folks that
would have preferred to simply dig-in and get to work, leaving the
debating for others.
As for the KDE stuff... as with you, I'm a KDE user, and knew as soon as I
saw the project favoring Gnome over KDE, that it wasn't something I'd be
interested in. However, apparently unlike many, I didn't go join the list
and argue the point. Rather, I simply wrote it off as something not worth
trying -- unless it developed a KDE version.
OTOH, given GNOME's very public targeting of the "simple" user, one who
gets confused by too many config options and the like, as well as the LGPL
licensing vs. Qt/KDE's GPL licensing, GNOME may have been better for the
"locked down corporate desktop" types as well as the "inhouse and
outsourced developed solutions" types (noting of course KDE's kiosk mode
and the fact that the GPL wouldn't prevent proprietary development if it
were to remain inhouse either, but GNOME will still seem a more "natural"
fit to the PHB types that were a prime UserLinux target, in any case).
That I /can/ admit. Such targets couldn't be farther from my own
interest, granted, the reason I wasn't personally interested for my own
use, but they'd be better for UserLinux's target market, it being what it
was.
In any case, it would seem that the community has yet another example of
how "design by committee" doesn't work so well. Design the specs first,
within a much smaller limited non-public group, or just let them evolve
naturally from the initial code, but in any case, get that initial code
OUT THERE, BEFORE the public announcement. THEN make the announcement,
and if the code and concept (or even just the code) are good, and
particularly if they uniquely fill a niche that no other product out there
fills (as arguably was the case for UserLinux when it was announced), the
users and further developers WILL come. The latter assumption, provide
even a rudimentary but unique solution filling a need, and they WILL come,
has after all been demonstrated time and again within the FLOSS community,
and is to a large degree what it's all about.
All IMO, FWIW...
Duncan
Show me the code! (UserLinux)
Posted Sep 16, 2005 9:18 UTC (Fri) by hingo (subscriber, #14792) [Link]
Coming from you, I'm truly humbled by such words.Anyway, once again thanks to lwn and the author for the article itself. This was so much more interesting than seeing screenshots of yet another distribution.
Show me the code!
Posted Sep 15, 2005 22:40 UTC (Thu) by gallir (subscriber, #5735) [Link]
I second your KDE analysis.And also deeply disagree the comparison of a desktop with SMTP or
database servers of the original article. Servers, especially those that
have a well known and widely used standard interface like SMPT, only
affect how programs interact. But GUI desktops affect how the _users_
interact with the computer. Those who chose just one desktop over the
other are the ones not thinking in users.
After reading the article I had the feeling that KDE "promoters" received
more blaming that deserved.
Show me the code!
Posted Sep 16, 2005 7:02 UTC (Fri) by frazier (subscriber, #3060) [Link]
But GUI desktops affect how the _users_ interact with the computer. Those who chose just one desktop over the other are the ones not thinking in users.Why? Please explain.
Show me the code!
Posted Sep 16, 2005 13:44 UTC (Fri) by gallir (subscriber, #5735) [Link]
A user wouldn't realise a change of theMTA (Postfix or Exim), but she does realise any change from Kmail to
others MUAs. The same, if not more, applies to the desktop environment:
it is another program.
Users don't care about the MTA running in their computers, but they do
care about the programs they use every day. By selecting one desktop you
are taking lot of decisions for the user, not for improving her/his
direct experiencie, but due to legal --licences?-- or technical reasons
--easier to maintain?--. Hence, adopting just one of the two major users'
desktop environment and simultaneously accusing KDE developers of taking
care only developers' interests is contradictory, to say the least.
Show me the code!
Posted Sep 16, 2005 17:12 UTC (Fri) by frazier (subscriber, #3060) [Link]
By selecting one desktop you are taking lot of decisions for the user, not for improving her/his direct experienceActually, making those decisions simplifies the experience, making it better for them.
It is important to remember that:
- Typically, it is easier to add software to system than to remove it, especially for less technical users. More technical users will also be more likely to add and customize to their desires.
- If someone wanted to add KDE (or GNOME if it were a KDE-only system), they could.
- The real growth for a business distribution on the desktop is people who are not currently using a Linux desktop at all, so the missing KDE (or GNOME if it were a KDE-only system) app won't be a big deal for most of the intended user base and furthermore those who will miss an app are more likely to be able to add it themselves. They are running a Linux desktop enough to miss an app, after all (unless it was somthing cross-platform like Firefox)
- Consistency: The dialogues among other things are different between KDE and GNOME.
- People maintaining systems are users too. More to maintain in the default is not advantageous because of:
- Security
- Support
- Help desk
- Training
Most people really don't want to use computers, they really just want to accomplish things and the computer is a tool they use. They don't particularily like them, they aren't an interest or a hobby, and they don't follow news for them more than they have to.
More software is not in the interest of the less technical user or businesses in general.
From the UserLinux perspective, look at the mission statement:
Provide businesses with freely available, high quality Linux operating systems accompanied by certifications,
Certifications are easier with a smaller set of software generate certification test for in regards to people and hardware (if applicable)
service, and support options
Easier to service and support software as a new employee or as an ISV if it's the same software you've been supporting in the past.
designed to encourage productivity
The idea tool is what you need and not much else. Though it is impossible to get a PC taylored to everyone perfectly, it is safe to say they don't typically need two browsers or two word processors, and the presence of both is more likely to confuse than aid.
and security
Less software to go wrong.
while reducing overall costs.
Streamlining will reduce costs, and the licensing used for a variety of the software (avoiding free/commercial dual licenses like MySQL uses) helps too.
Show me the code!
Posted Sep 25, 2005 22:47 UTC (Sun) by dkite (guest, #4577) [Link]
Great ideas. It didn't work.UserLinux made a so basic mistake, a fundamental misunderstanding of FOSS
that surprised me considering the reputation and background of it's
backers.
Free software is only about users when the users can provide a
contribution.
UserLinux started off by alienating at least 1/2 of the developer base of
free software. UserLinux depended on contribution. Very bad start.
UserLinux attempted to define the free software user experience by
excluding worthy projects and their developers. Bad idea.
It was very predictable. Anyone who used or contributed or thought highly
of any of the excluded packages were uninterested. Not only uninterested,
but actively excluded by comments similar to yours. So the project died.
Ubuntu could afford to pick favorites because they hired the help, and
didn't depend on contributions.
I for one, who contributes substantially to FOSS, was insulted by the
whole presentation and philosophy behind UserLinux. It wasn't for me, and
actively questioned the morality of my contributions. So I, along with
many others, didn't contribute.
Derek