On Wed, 13 Feb 2002 15:57:07 +0700, Brian Durant <[EMAIL PROTECTED]> wrote:
> >You trivialise the issue. Drivers are supposed to be written by hardware
> >manufacturers (who actually know what the hardware is about), not by
> >software/OS designers (who must reverse engineer the hardware to know
> how it works).
> 
> Yes, you are of course correct in this, but I don't feel that I am
> trivialising the issue. The reality, for someone like myself, is that we
> in principle would gladly buy into the Open Source concept, but we want
> something that will work. Not necessarily out of the box, but something
> that is not "voodoo" for 99% percent of newcommers to the OS. You don't
> do that in the Mac OS and you don't do that in Windows.

Is hardware support really as bad as you seem to make it sound? I've loaded
various versions of Mandrake on a variety of different systems, and I've never
had any problems that I couldn't easily fix.

> >Manufacturers are reluctant to write Linux drivers, forcing the community to
> >come up with their own. Why is this the case? First and foremost, there is no
> >commercial incentive to support an OS that only has a few percentage
> points of
> >the desktop market. This problem applies to all but one x86 OS.
> Secondly, many
> >manufacturers misunderstand how GNU/Linux works, believing it is somehow
> >'viral' (to use MS terminology). To them, releasing drivers means
> letting their
> >intellectual property secrets out into the open, which elimiates any
> >competitive edge that company may have had.
> 
> This is really where the problems start to come to light. Apple, while
> dependent on 3rd. party companies to provide certain hardware (and
> software), largely controls both the hardware specs and the UI, making a
> far more homogenous package. M$ leverages hardware as well due to its
> large installation base and in some cases has produced or has had
> hardware produced under the M$ name to M$ specs, sometimes in a duopoly
> with Intel. This is also where the Linux business model is partially
> faulty. Linux is too dependent on hardware manufacturers, as well as
> splitting efforts too often between, similar GUI or programs. A viable OS
> is not just based on hardware or software, but a combination of both.

Is Linux really "too dependent on hardware manufacturers"? After all, there are
open source drivers for most major pieces of hardware, developed independently
of the manufacturers. Even ATI and Nvidia video cards have open source drivers
(in addition to proprietary ones from the companies themselves).

> Linux has not succeeded in leveraging hardware developers to any
> meaningful extent as far as I am concerned.

GNU/Linux runs on almost any hardware platform imaginable, from tiny embedded
devices to mainframes and supercomputers. Windows and MacOS can't do this, and
for the most part they are tied to x86 and PPC respectively. GNU/Linux may not
have much desktop market share, but it is incredibly strong in other areas (e.g.
servers). IBM, for example, are adopting it for its entire hardware line,
literally ranging from wristwatches to supercomputers. Sun, Compaq and HP are
adopting it as well, and they all have pumped significant amounts of money into
open source development. Last year, IBM spent about $US1 billion on GNU/Linux,
and in that same period they recouped almost all of it. If that isn't
"leveraging hardware developers" then I don't know what is.

> >The reality is somewhat different: Linux (i.e. the kernel) is licensed
> under a modified GPL which allows proprietary binary-only modules.
> Companies like Nvidia have taken advantage of this, and
> >have released very capable drivers. 3dfx and Matrox went one step further by
> >openly co-operating with open source hackers to produce open drivers.
> >
> >Why does Windows seemingly have such great hardware support? Because it
> >has over 90% of the desktop OS market, it cannot be ignored by hardware
> >manufacturers. Do you really think that MS write their own hardware drivers?
> 
> This does not ignore the fact that most Linux advocacy is done in the
> area of trying to get hardware producers  to support open specifications
> for Linux drivers or to use Linux in the embedded chip market. What
> prevents a dedicated Linux group from producing a sound card, video card,
> etc. that is made for the Linux market? The same goes for PDA's. Sharp's
> new Zaurus prototype using Linux looks great, but who says that others
> can't build a better mouse trap? Is it maybe because the market isn't
> there yet? I'm not sure, but Linux advocacy reminds me a lot of the Mac
> advocacy that I saw some years ago, when Apple kept losing market share.
> Lastly, coding for Linux projects, whether they are for the kernel or
> specific software such as drivers or other things can easily be shared on
> the Internet. If someone drops out or burns out, there is usually someone
> else that joins the team and helps out. Physical production can't be
> shared on the Internet for obvious reasons (Beam Me Up Scotty)!, it
> requires financial investment, etc. Another kind of commitment is
> required. All of this is obvious to everyone, yet it seems to me that it
> is here where Linux' weakest link exists.

I'm not sure if I'm understanding you correctly, but are you suggesting that the
open source community make their own hardware products from scratch? In most
cases, this is far more trouble than it's worth, and we'd be better off simply
lobbying manufacturers to support GNU/Linux. There are some products made
specifically to run GNU/Linux, like the Zaurus (as you have mentioned). The
closest thing to your proposal would be the Simputer (http://www.simputer.org).
It would be economically unfeasible (and unnecessary) for a group of people to
join together and decide to produce, say, a sound card. Even if it were
possible, they would never reach sufficient economies of scale to make a decent
return on the investment. There is more to a company than technical expertise,
with or without external funding. It would be much simpler to make GNU/Linux
drivers for an ordinary device. Why should I buy a new graphics card simply so I
can use GNU/Linux? I want my existing hardware to work, as do most other people
(particularly newbies).

> >Apple are in a similar situation. They have their own little hardware
> market, of which they
> >would have about 99% share (with Darwin and GNU/Linux making up much of the
> >remainder). They make their own boxes, giving them ultimate control over the
> >entire platform. Consequently, hardware designed for Macs works exceptionally
> >well (often better than how they work in Windows).
> 
> In reality, I don't think that we disagree as such, except maybe with
> regards to how user friendly the Linux experience should be for newbies.
> This discussion has been fruitful for me, if for no other reason, that I
> now know that I should look more closely at Nvidia, 3dfx and Matrox video
> cards ;-) Finally, I think that Mandrake could improve their site for
> people doing research about what video card/sound card to buy, by
> providing a text file or PDF that provides a list of all of the drivers
> for recognized cards that are available at install for each version i.e.
> what cards are recognized when installing Mandrake 8, 8.1, 8.2, etc. on a
> new system. While the categories of "known" and "tested" hardware are
> useful, it really doesn't provide the whole picture.

There are few reasons to install Mandrake 8.0 when 8.1 is the latest version. I
agree that a downloadable list might be nice, though. There are also plenty of
GNU/Linux sites that have lists of supported hardware and descriptions on how to
get particular pieces of hardware to work.

-- 
Sridhar Dhanapalan

"Maybe somebody else comes up with a better way to do it, or with a really
compelling reason to. "Feel free to try" is definitely the open source motto."         
 --
Linus Torvalds

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Reply via email to