OSDL and the Linux kernel community

OSDL and the Linux kernel community

[Otto Wyss]

I've just read about Greg Kroah-Hartman (one of the 17 kernel 
developers) proposal about what OSDL could do to improve the kernel 
work. As a none-kernel developer I think it's important to look closer 
at some of the issues.

First I have to mention that the kernel is an important but only _one_ 
part of a Linux system a user may buy in a shop. There are many more 
parts which have to fit in until a system is bought. IMO the OSDL's task 
is to care for all of these parts, not only at the kernel. Nobody will 
buy a computer if it only has a superb kernel but the rest doesn't fit 
equally well.

Second the breakthrough of the Linux desktop system has been predicted 
so many years, it's hard to find the correct number. It's not worth to 
discuss this much, just think the OSDL thought it necessary to start a 
survey about it last year. The results of this survey can be read here 
(http://www.osdl.org/dtl/DTL_Survey_Report_Nov2005.pdf). Besides the 
usual points as e.g. hardware difficulties, there is one single point 
which stands out. Doesn't it strike you odd when you read that these 
users still ask for applications not available on a Linux system? Even 
if the majority had Linux deployed? Can you imagine what the first top 
inhibitor of the Linux desktop adoption means?

Well the answers is quite simple, it means a low market share! A market 
share so low that manufactures don't care about providing drivers, even 
so low they don't even provide documentation. Too low that vendors don't 
care for using parts which have Linux drivers, let alone selling Linux 
systems even if it's cheaper for them. Don't you think it is much more 
important to raise the market share above a minimal necessary level?

So the number one priority to solve many of the problems of the Linux 
kernel is to increase the market share above this minimal level. To fix 
the first top inhibitor of the Linux desktop adoption. I guess very few 
of you have an idea how this can be achieved. Neither had the desktop 
architects at ODSL before I mentioned it to them. Even then they are 
reluctant to listen to me (see 
http://lists.osdl.org/pipermail/desktop_architects/2005-December/000349.html) 
albeit they finally seem to agree (see 
http://lists.osdl.org/pipermail/desktop_architects/2005-December/000369.html). 
At least they and most probably you don’t know any alternative.

So if Linux should finally become widespread used by everybody my 
“cross-platform development” initiative (see 
http://wyoguide.sourceforge.net/index.php?page=Cross-platform.html) has 
to succeed. Only then becomes a Linux system attractive for the ordinary 
users and therefore for the vendors and the manufactures.

You may ask now why do I tell this here and wonder what you have to do 
with. Well first it was Greg Kroah-Hartman proposal to the OSDL and 
second because I think my proposal will solve many of the limitations 
you currently face in your work. I think you are well advised to 
advocate this to the OSDL so they finally look for a desktop architect 
in their fellowship program 
<http://www.osdl.org/about_osdl/Maintainers/document_view>. You may 
further advocate the use of the wyoGuide guidelines so the Linux desktop 
only has just one set of guidelines instead of several. And you are free 
to participate in wyoGuide so it’s not only a good but the best for the 
Linux desktop.


O. Wyss

-- 
Application guidelines: http://wyoguide.sf.net/

Re: OSDL and the Linux kernel community

[Nick]

On 28/01/06, Otto Wyss <otto.wyss@orpatec.ch> wrote:
> I've just read about Greg Kroah-Hartman (one of the 17 kernel
> developers) proposal about what OSDL could do to improve the kernel
> work. As a none-kernel developer I think it's important to look closer
> at some of the issues.

Of course, you can do what I do at work - promote GNU/Linux in _every_
conversation with the users (no matter waht the topic).

Only this week I recovered 1.2GB of pictures from a users home
computer that die running eXtra Problems.

Using Knoppix and a USB stick with the bad drive mounted on hdb (ok,
took 6 'runs' to recover all data) , he was WELL impressed (his wife
even more so...) and started asking questions...

Nick

Re: OSDL and the Linux kernel community

[Chase Venters]

On Saturday 28 January 2006 09:33, Otto Wyss wrote:
> First I have to mention that the kernel is an important but only _one_
> part of a Linux system a user may buy in a shop. There are many more
> parts which have to fit in until a system is bought. IMO the OSDL's task
> is to care for all of these parts, not only at the kernel. Nobody will
> buy a computer if it only has a superb kernel but the rest doesn't fit
> equally well.

Yes, but if the kernel doesn't support the user's wireless chipset and all 
they can get is binary blob drivers for their graphics cards, the rest of the 
system will seem naturally worse.

> Well the answers is quite simple, it means a low market share! A market
> share so low that manufactures don't care about providing drivers, even
> so low they don't even provide documentation. Too low that vendors don't
> care for using parts which have Linux drivers, let alone selling Linux
> systems even if it's cheaper for them. Don't you think it is much more
> important to raise the market share above a minimal necessary level?

This problem exists, but even ATI and NVIDIA both recognize that there is a 
market for linux users. So they provide (some) level of effort to write 
drivers. It's probably their concerns about IP and the IP climate that keep 
their drivers closed.

> So if Linux should finally become widespread used by everybody my
> “cross-platform development” initiative (see
> http://wyoguide.sourceforge.net/index.php?page=Cross-platform.html) has
> to succeed. Only then becomes a Linux system attractive for the ordinary
> users and therefore for the vendors and the manufactures.

I don't see how cross-platform development makes Linux attractive to 
"ordinary" (non-developer) users. I don't think asking everyone to swallow 
some red pill is going to make anything any bit better.

Cross platform toolkits like GTK and Qt already exist. Last I heard, Trolltech 
even had some large customers (Adobe, etc), so who knows - maybe Photoshop on 
X isn't so long off?

> You may ask now why do I tell this here and wonder what you have to do
> with. Well first it was Greg Kroah-Hartman proposal to the OSDL and
> second because I think my proposal will solve many of the limitations
> you currently face in your work. I think you are well advised to
> advocate this to the OSDL so they finally look for a desktop architect
> in their fellowship program

The limitations kernel developers currently face is in the availability of 
documentation and the cooperation of manufacturers. While the rise of Linux 
adoption will help this, it's up to the desktop people (read: desktop 
architects) to worry about desktop stuff. I'm a subscriber there too, and I 
realize you didn't feel all that at home with your guidelines, but don't you 
think LKML is the wrong place to be airing them?

I read Greg's words on OSDL too, and it seems (from where I stand, anyways) 
that things are going well in that area. But that initiative is focused on 
the kernel and kernel developers, and yours is focused on desktop.

> O. Wyss

Cheers,
Chase

Re: OSDL and the Linux kernel community

[Greg KH]

On Sat, Jan 28, 2006 at 03:16:06PM -0600, Chase Venters wrote:
> 
> I read Greg's words on OSDL too, and it seems (from where I stand, anyways) 
> that things are going well in that area. 

Thanks, I hope so.  Only time will tell :)

> But that initiative is focused on the kernel and kernel developers,
> and yours is focused on desktop.

Exactly.  Desktop developers already have their own foundatations and
orginizations where discussing desktop things, like tool kits, are best
brought up.  This is a kernel specific mailing list and initiative with
OSDL.

thanks,

greg k-h