[Sussex] Linux Heavyweights Sound Off At Summit

Steve Dobson steve at dobson.org
Wed Feb 9 20:57:01 UTC 2005


Angelo / Geoff

Isn't it nice to have some good threads to think about back again.

On Wed, Feb 09, 2005 at 06:22:33PM +0000, Geoffrey Teale wrote:
> On Wed, 2005-02-09 at 17:31 +0000, Angelo Servini wrote:
> > IMHO is, its time maybe for the Open Source community to address
> > this.  Im all for freedom of choice, but - how fast would a unified
> > desktop advance, if ALL the desktop teams pooled their efforts?
> 
> See www.freedesktop.org - we're all singing from the same prayer sheet,
> but we all sing slightly differently.

And long may we all sing differently.

The last thing I want is to be forced to do things "a different way"
just to get a large bunch of id10ts to adopt Linux.  Most computer
users don't care, and will never care, how their OS and apps were 
developed.  Just like most people just want a car to get them from
A to B, or cheap, battery grown chicken meat, or ... pick any issue
and there is a group that campaigning for the status quo to be changed.

I happen to like Gnome ATM, but I have changed from Enlightenment and
before that FVWM.  I tried KDE, but found I like Gnome more to my
liking.  I don't want to have to use KDE because IBM, HP, Novell,
RedHat & SuSE got together and decided that they would standardize on
KDE just to beat Windows.

The Free DeskTop is the way to do things: define standards not
implementation. I am then free to cherry pick the applications that
are best for me from which ever desktop or task area that I need to use.
 
> > I read a mail by someone on the ubuntu forums who sounded off to
> > ATI for their crap support of Linux. But, who can blame them.  
> 
> I can.  

And I.  And Alan Cox is another - and he does carry some weight in
this community.
 
> They refuse to release details of how to use the hardware they sell
> people and they justify that by saying "it's OK we'll give you some
> closed software to let you use the hardware you buy" - thus people
> intending to use a Linux system buy ATI hardware because ATI say it will
> work and then find out that the ATI drivers basically don't work at all
> reliably in any sense of the word.  Frankly this situation is worse than
> ATI not providing drivers at all - they are giving lip service to Linux
> purely to try and get their foot in a market that they can't be bothered
> to approach sensibly.
<snip>

I had the luck to be sitting next to Alan C, after a Linux show a few 
years back, drinking (he, or rather RedHat, where buying too).  I only
name drop so you can see that this thought came from a greater man
than me.  He pointed out that companies that keep their hardware closed
are not considered when the kernel developers plan the next kernel
developments.  How can they be?

Alan's point was never to buy any hardware from a company than hasn't
opened up it's hardware specs.  Any driver that is in the kernel will
be supported by the kernel developers themselves.  Who would you rather
code the driver for your graphics card?  A kernel developer, who's work
is checked by many eyeballs, or a paid employee, who has a PHB standing
over him shouting:  "We have to release this code tomorrow"?
 
> > They perceive the Linux community as a small insignificant group
> > and M$ is king.  
> 
> For consumer level PC's this is almost definitely the case right now.
> Bear in mind though that Linux has a bigger share of the desktop than
> Apple does right now and Apple kit runs ATI and Nvidia hardware.  The
> big difference is Apple can sign an NDA with ATI and write their own
> drivers.

I disagree.  There is a new market for cheap computer components just 
around the corner - and in that market place cost is everything.  The
market is the PC as a consumer electronics component.  I found this

              http://www.plusworld.co.uk/Bigissue02.htm

There are around 5 million Sky customers in the UK.  If half of them
convert to Sky+ that £300 million/year to sky!  Sky will sell the
box at (or below cost) because the licensing is worth so much more.

Go into any Dixons and ask the senior sales staff what is the key
feature of any of their product ranges and they will respond: Price.
It there are two units with the same form, fit and function the one
that is cheaper will sell more.

Linux *WILL* dominate in this market space.  So long as M$ charges a
license fee for each unit sold (and how can they not) that cost will
be passed on to the consumer, and that will effect the price you
see in the shop.

All that we are waiting for is for the price for a PC to hit 
community price levels - £200 - £300.  When that happen PC
will sky rocket again - and a company shipping Linux on those PCs
(like Walmart in the US) will be able to offer them for less.
 
> > They will make a wrong and uninformed choice, and pool all their
> > efforts into supporting the majority OS, as seen by their eyes.  

The masses will always make wrong and uniformed choices - because
the masses are generally ignorant of the facts (and I include myself
in that statement for subject I know little about - like healtcare).

For example looking at the number of drivers who pick a 4x4 for the
family car.  Does it matter to them that the car is more susceptible
to side wind?  Does it matter to them that the bull bars they fit
are more likely to kill than to injure should then hit someone?  Does
it matter that the engine, having to push that large un-aerodynamic
brick, is pumping out more of the gasses that are killing the 
environment?  No, they just think the car "looks cool", or gives
them a false sense of safety.
 
> > If the Hardware community on the other hand perceive Linux as
> > becoming an up and coming better bet.  Then you will see them burst
> > with activity to provide support.  Its all driven by profit.  
> 
> Yup.  
> 
> One of the biggest problems with consumer grade hardware is that almost
> all of the functionality lives in software these days - Winmodems,
> scanners and printers all suffer heavily from this.
> 
> The problem is software development is cheaper than hardware development

Of course S/W is cheaper than hardware.  Each version of H/W costs a
lot of money to design and get right.  If a H/W engineer gets it wrong
and sends 5v down the wrong wire on the motherboard then bye-bye CPU.

If a s/w engineer gets it wrong what does (s)he do?  Puts debug print
statements into the wrong code and runs it again to see how it failed.
And (s)he keeps doing that until (s)he figures it out!

That is why h/w engineers hate s/w engineers :-)

> (so there's more profit) but it's also more transient - a lot of cheap
> modems, scanners and printers that came onto the market five years ago
> simply won't work with Windows XP never mind Linux - that's because the
> companies that made them put all the processing in software and didn't
> maintain it beyond the shelf life of the product.
>
> Don't think that drivers are purely a linux problem - they're
> increasingly a problem for all platforms only masked by the fact that
> consumers rarely upgrade their OS without upgrading their kit (I know
> several people still running Win95 and win98 !).

I see nothing wrong with this.  If the power is moved into software then
the hardware becomes cheaper.  As software manufacturing cost are now 
(effectively) zero, then I can afford to get better hardware more often.
 
> > If linux is to be king, then we need to win the Desktop battle. Thats
> > my 2 denarius, anyhow.
> 
> It's coming, but it's slow.  Home desktops are the last place Linux will
> take hold - it'll get there if and only if it becomes dominant of
> business desktops.

Business will switch to Linux because it will save them money.  Nik B
has one client of his that saved £48k my switching to Linux - and that
is just the cost in licensing.  How much more have they saved in not 
loosing productivity to the various virus attacks, etc.

Companies that have switched will find they have more funds available 
for other areas of business, like more staff or more R&D.  This will
help them grow their business, and make them more successful (if they
get the other bits of doing business right too) than their Windows using
competitors.

I don't worry about the business market any more, it will switch - Linux
has already passed the turning point.

>                     Even if it does end up on most business desktops it
> may not drive further adoption because Linux apps work to open standards
> and don't preclude interaction with other software (the opposite of
> Microsoft's approach which has allowed them to leverage an initial
> monopoly of business desktops to gain monopolies in other markets).

Geoff: Don't you mean "...it *will* drive further adoption..."?
 
> If anything the key battle is not getting open source everywhere but
> getting open standards everywhere.  Here there is more hope.  It seems
> like the worlds legislators are beginning to realise that there's no
> point retaining records for 7 years if you have no way to understand the
> file at the end of that period.

Agreed.  See my link above.

I would finally add that I don't want Linux to change to be a Windows clone
to court the Window users.  Linux is, and more importantly *must remain*,
different to Window.  While Winows may be in competition with Linux; I don't
think Linux (as a whole) is in competition with Windows.  IBM, HP, Novell,
RedHat, SuSe, ... may be in competition with M$, and using Linux to do it,
but that is not the same thing.

Linux has gained many, many advances by different groups wanting to 
take Linux in different directions.  We got XFS, JFS, and other enterprise
quality features because of the way SGI and IBM wanted to use Linux.
Linux has taken over the super computer market (but it was written for 
a 486 for God's sake!!!!).

But I like the Real Time (RT) groups contribution best.  IIRC Linus,
when he first saw the work of the RT group, said:  (not a direct quote,
but one from memory, mine - so no parity checking :-) "I like the work,
but real time is not an interest of mine, so I be unlikely to accept any
of the RT patches in to the main Linux tree".  Then they came up with
the Preemptible Kernel patch.  Here is what the Linux Kernel Config
Help text (in the main Linux tree) says:

    This option reduces the latency of the kernel when reacting to
    real-time or interactive events by allowing a low priority process to
    be preempted even if it is in kernel mode executing a system call.
    This allows applications to run more reliably even when the system is
    under load.

    Say Y here if you are building a kernel for a desktop, embedded
    or real-time system.  Say N if you are unsure.

Linus had to eat his words.  He's had to do this several time, as he freely
admits.  That's the power of the FOSS development model, sometimes tangents
throw up features useful to a wider audience.

If we focus too much on getting the desktop market we could loss so much.
If we focus too much on the needs of the desktop then we (or rather the
kernel developers) could start making decisions that cause problems in 
other areas, just to please the desktop people - like putting the GUI 
functionality into the kernel - and that way lies dragons.

Steve




More information about the Sussex mailing list