Re: Xandros, Lindows, Debian, the future of distros

[Jeff Licquia <jeff@licquia.org>]

On Sun, 2003-01-19 at 12:13, mike808@users.sourceforge.net wrote:
> > ... CrossOver isn't perfect (which it isn't).  They got pretty embarassed 
> > with their claims of Windows compatibility, ...
> 
> I can understand that. So why didn't Codeweavers take any of the heat?
> Or Lindows offer to "incent" Codeweavers to meet their [Lindows'] claims 
> (that apparently went beyond what Codeweavers delivers)? Any embarassment
> was mostly of their own making and marketing, not Codeweavers'.

Well, right.  I wasn't implying that Lindows' embarassment was the fault
of CodeWeavers.  CW has been pretty forthcoming about the limitations of
their product, which I respect quite a bit.

> >> credit for putting "yet another pretty face" on the Debian install.
> > 
> > Well, so did we (Progeny).  Except that we didn't keep our pretty face
> > to ourselves; we gave it back to Debian, just as we promised we'd do. 
> > And you can use that pretty face to install real Debian today, without
> > licensing fees, or EULAs, or any other behind-the-scenes shenanigans to
> > distract you.
> 
> Hmmm. I might be heading to that URL soon to check it out.

http://hackers.progeny.com/pgi, if you're wondering.  You can download
images from there, or make your own customized installer.

> > We've come a long way to help Debian forget Corel's behavior, I think.
> 
> And Corel is gone (from the Linux business). Why blame Xandros for picking
> up the Open Source pieces and forging a new business out of it?

I don't object to that.

> Just
> as what will likely happen with Mandrake. The company may die, but the
> code doesn't.

I hope it doesn't with Mandrake, but you never know.

If most of the work Xandros has put into their distro is proprietary,
then yes, their code will die with the company (assuming it dies).

> Unless you have some info that the core Xandros team is
> mostly the former Corel Linux team, and has the same mentality.

Which I might.

> > [click'n'run]
> >
> > That (the beta) was something they were grilled on pretty heavily when
> > they met with a bunch of us.  My conclusion from that conversation was
> > that Lindows was naive; they didn't understand the problems, and just
> > did what companies normally do with betas.  They promised to do better,
> > which they seem to have done up until recently.  Single stupid event, or
> > the beginning of a trend?  Time will tell, I suppose.
> 
> But, from a single "stupid event" of Corel's lack of credit to Debian,
> you're not willing to even consider that they were also, perhaps, naive.
> You've concluded that Xandros must suffer from their Corel ancestry, and
> are willing to entertain justifications and explanations from Lindows, 
> but not from Xandros? At least you make clear your bias, for the reader
> to take into account.

I didn't claim that my bias is unjustified, just that you might have to
treat it that way.  Neither did I claim to not have good reasons, just
that, if I have any, I'm not posting them to public mailing lists.

Now, I don't claim any special bias towards Lindows.  Their recent
actions have me plenty disturbed, especially seeing how Bruce Perens has
reacted.  But I still give credit where credit is due, and they are due
some credit for their positive engagement of Debian in the past.

> > As for Click-N-Run, who knows?  Maybe it'll work, maybe not.
> 
> Xandros is moving, I think to a similar model with their "Xandros Network".
> And all of these are based on the "Red Hat Network" model of subscription
> access to software update repositories. Also, the boys from Redmond are
> also on this plan as well, they just don't call it a subscription (yet).

Microsoft's subscription model, as I've heard it described, is a little
different from RHN, Click-N-Run, etc.  The various Linux distributors
let you use the software you've downloaded after your subscription runs
out; you're just on your own regarding updates.  MS, on the other hand,
wants Office to shut off if you stop feeding the meter.

> The value will come down to how well they will execute the "tweaks" required
> to integrate package updates with their particular distribution. I think
> a lot of energy is going into figuring out how to implement a "rolling 
> release" type of distribution. The market pressures are moving there, and
> we see this in the announcements of shorter release cycles. First annual,
> and quickly moving to 6 months. A key problem is the broadband vs. CD
> delivery mechanism.

I *really* like how you say that. :-)

> > BTW, the price for the distro includes a year of Click-N-Run, so it's
> > hard to fault them for shipping a stripped-down base if you can install
> > add-ons for free.
> 
> Which totally sucks if you don't have broadband. It's hard to remember
> the vast majority of folks that don't have it when you've gotten *very*
> used to it. It's quite seductive.

*shrug* Some folks are impossible to please, I guess.

Adding a CD to a distro package costs real money.  I can see why they'd
be keen on reducing the disk count, especially if they can provide a
service to replace the disks.  As a customer, I'd sure rather run long
downloads than pay $$$ per CD extra upfront for software I might never
use.  YMMV, of course.

> > > Does anyone have first-hand experience with Lindows and care to comment?
> > 
> > I've seen it running and talked to their developers about it.  In
> > particular, apt still works, so you don't have to use Click-N-Run if
> > you're comfortable with apt.  It looks slick, and appears to be "user
> > friendly" as far as that goes.
> 
> Sounds about the same as Xandros. You can also opt-out and add your own
> apt repositories. The GUI tools to do this are nice, and easy to understand
> for desktop users (i.e. no using 'vi' or 'emacs' to edit /etc/apt/ files).
> 
> As mentioned, the "Xandros Networks" service looks to be similar to 
> Click'n'Run, but they haven't worked out the pricing model yet.

Yup.  I wonder how long it'll take for Lycoris, Mandrake, etc. to do the
same.

> That said, since all of the distros have "online update" in some form
> or another, and are increasingly looking to that as a revenue generator
> (since bandwidth is cheaper and faster for just-in-time delivery of the
> "inventory" (aka software) than CDs and physical media), especially for
> its recurring nature, as well as not having to share costs with distributors.

Described exactly that way, I think any such effort is doomed.  Apt is
just too easy.

The only way to make this work is to put a spin on it that makes it
better than apt alone.  Click-N-Run, for example, is all set to
distribute proprietary software as well as free software, since they
have full control over the delivery mechanism.  I'm sure Xandros is
looking at the same angle.

> So, is the future something like a floppy install and broadband required?
> You *know* Micro$oft is drooling to get this, if they think they could get 
> away with it. That's exactly what "product activation" is all about. However, 
> they have a problem in that their proprietary licensing model has to manage 
> who is allowed to have what. Yet another reason to love Open Source.

Network installs are a good idea, period.  It's indispensable for secure
installations, for example.  Without installing the security updates as
a part of the installation process, you're always going to have a race
condition between the time the install finishes and the time the updates
get installed.

The problem with networked installation is that it's hard to control. 
If you make your programs available over the net, people will find ways
to get access to them.  Of course, in the open source/free software
worlds, this isn't a problem.  This is why, for example, Debian has been
the market leader in software distribution over the network; everyone
else has too many vested interests in trying to lock people out, so
they're scared to try new things.

> But, this is also a problem for the Open Source repository folks since they
> are effectively competing on price for the service, since they can't 
> charge for or restrict what you download from them. And since the commodity
> in that model is the bandwidth, it's they who make the money, not the
> repository managers. It's going to be tough convincing the public on the
> value and relative merits (to your competition) of your "packaging" skills
> if you're a distro.

I couldn't agree more, though I'm a lot more optimistic on the
prognosis.

> We're seeing signs of this in the proliferation of "targeted" distros - 
> Firewalls, Enterprise, RealTime, Embedded, Desktops, etc.. In that world,
> Debian does, I think, have the right approach in focusing on the packaging.
> Just as the BSD folks have also focused on that to some extent.

People are just now waking up to the advantages of freedom.  It will be
interesting to see how closed-source vendors react.


-
To unsubscribe, send email to majordomo@luci.org with
"unsubscribe luci-discuss" in the body.