Re: hardware compatability

[ABrady <kcsmart kc rr com>]

On Tue, 15 May 2001 08:41:48 -0700 bob jones <bobj eskimo com> spake unto
us:

I only skimmed over the original because, of what I read, I agreed with
the central point and disagreed with what some of it _appeared_ to say,
but didn't want to expend the effort on arguing about it. I mostly didn't
mess with replies to it either. Now, I guess, I can get into it. And if
some of it is redunant, I apologize. Slap me, or send me to /dev/null or
something.

> Hi folks, and thanks for your responses. You have made several
> suggestions: not purchasing unsupported hardware, supplying the needed
> code myself, being patient, and being thankful for Linux as it is. There
> were also some constructive suggestions and links to check out, and I
> appreciated that. However, some of you seem to feel that I am attacking
> Linux. That's wrong; I'm not.
> 
> After reading all of the responses, I'm still certain that I'm on solid
> ground in stating that a serious Linux weakness is the lack of drivers,
> the fact that some existing drivers do not behave correctly, and that
> these factors limit hardware and activities that most of us would want
> to pursue using Linux in preference to other OS's.

I couldn't agree more. But, that weakness is going to persist until a few
things happen. Manufacturers are going to have to feel charitable and open
up their source. And/or people are going to have to be willing to pay
_in_significant_numbers_ for manufacturers to make driver development
affordable to their bottom line. And/or someone will need the product
themselves enough to justify the effort in engineering the drivers. And/or
enough people are using the product to justify expended effort in working
on the project.

I see the argument given from time-to-time "don't buy their product and
write to tell them why you don't." That sounds wonderful. But, producers
aren't going to be nudged to cooperate simply because less than 1% (that's
usually the case) write to complain _and_ will actually spend some money
once the effort is finished. 5-10% (very generous here) will write or
complain in some manner. In the end, far fewer will actually part with the
money that makes it worthwhile for the manufacturer to spend time creating
drivers.

Opening the source is a good idea. But, they own the product and have the
right to open it or not open it. It doesn't matter what is the best method
or anything else. The owner still has absolute right to do with the
product, that they created, anything they wish to do with it.

The fiasco with tulip modules (ala Linksys, but some others as well) is
exactly the type of problem being mentioned. But, that is the fault of the
maker of the product, not linux. Nobody on earth (including the makers of
the boards themselves) is able to make a one-size-fits-all driver because
manufacturers change things and break what used to work. Add to that the
fact that source isn't released, leaving someone coming up with a method
that works, but may leave something important out. Get a new card and
whatever was missing in the product that worked before may now become
apparent. Fixing the new one tends to break the old one in many cases.

The best solution is not announcing a boycott (which isn't likely to work
anyway). Rather just don't buy the product. Fine, if it makes one feel
better writing to someone and telling them why you won't buy the product,
feel free. It isn't completely out of the question that they might open up
the source or start producing drivers. I'm just saying, don't think that
_your_email_ is going to make them see the light, because if that's what
it's going to take to make you happy, you're going to be unhappy for a
long time. Maybe forever.

> As for checking out hardware compatibility, I always do that, on the HCL
> and with the RedHat-list. The problem is that sometimes hardware drivers
> work, and are somewhat supported, but only work partially. That seems to
> be the case for the HP970Cse, which prompted my posting in the first
> place. It isn't the only driver problem that is out there, as we know.
> 
> Back to the real core of the problem that I posted about, and how most
> of us cope with it.
> 
> I'll bet that most of you run dual boot systems, and that MSW gets a
> fair share of the action. Many of you no doubt must run MSW as an
> employment requirement. Six years ago I did too. Now I'm fortunate to
> need to run Linux only. Those of you who run MSW have access to two OS's
> -- and
> can run each OS to do whatever task it is best at. Therein lies a simple
> fact: to make use of what computers can do today, at home and at work,
> you need two computers, or two OS's. I'm not writing about games here,
> but about real and productive interests that people have, such as
> imaging,
> printing in color, scanning, music, etc. Ideally, one computer and one
> OS
> would be sufficient for such a wide range of applications, and we would
> want it to be Lunux on a PC. That's what I posted about, not to
> criticize Linux.

I have 'Doze 95 (for my 2-year-old) for games. It doesn't get used for
anything serious. It's also not in dual-boot config, doesn't have a
printer, has no access to the internet (I can get into it via samba, but
it can't get out), etc. It's used for games and learning (because these
are 2 more areas where linux sucks) and that's all. I expect it to crash
(and it doesn't disappoint me) and I expect to have to reinstall
everything occasionally (still no disappointment here). It serves it's
purpose. But, I'm not about to try putting it to any kind of productive
use because I worked with it for a long time and already know almost all
of its failings.

My other 3 computers are a MAC (OS9), a linux box and one with NT4.0
server (another I won't actually try doing anything serious with) to learn
about it for work, trying to fix things our "computer experts" aren't able
to figure out. The NT box has a single purpose and will go back to being
a linux box when it's purpose is finished. None of these dual-boot either.

That means, if a device doesn't work in linux, I never have the chance to
compare it with any 'Doze stuff because those are toys and are limited on
the hardware side. Still, I can see drivers are a major problem in linux.
It isn't the fault of linux, but that's the way it is.

But, the fix isn't likely to be in until either manufacturers are
convinced they have a market ($) or they simply feel kind-hearted enough
that they open up the source.

> I have an unreasoning and, probably an uninformed, distaste for MS,
> for MSW, and for all that it represents. I won't buy it and its apps,
> and I won't run them. Therefore I have a problem when some of the things
> my computer should support don't work. My fault for being stubborn, but
> I won't dual boot to MSW to solve my problems.
> 
> I wish more of you, and more of the computer-using population, would
> also
> refuse to run MSW and its apps, at least at home. That might, in time,
> solve many of the problems that Linux has.

Sorry, but until I can find enough on linux (or even MAC, plus the price
of MACs come down) I'm going to use it for games and education. Nothing
useful on an adult scale, though, because it isn't stable enough.
 
> I was interested to read in one response stating that Linux is moving
> away from a mainly server focused role. I didn't know that.

That's true, if you believe what the gnome and kde people are preaching. I
just hope we don't end up with M$-linux (i.e. instability and bloat on top
of an underlying unix environment). They both seem to be determined to
head us in that direction.

> In response to the suggestion that I write the software that is missing
> and needed, I have to say that a little knowledge of what is involved in
> writing drivers would have eliminated that suggestion before it got past
> the thinking stage. I won't list reasons for this, but most of you know
> them anyway. They are very solid reasons, and apply even to experienced
> programmers.
> 
> In response to the suggestion that I be grateful to have the Linux OS at
> all, I have to say that I am very, very grateful. Linux has made using
> computers important to me and added measurably to my life. No other OS
> could have done that.

Wish I had seen that. What a silly comment that had to have been!

I should be grateful that _any_ printer or network card or scanner or
mouse, etc works in linux? Really a useful attitude, that one!

> Back to the main problem, which is that to make full use of your
> computer, you need two OS's, or two computers. Only Linux can do what
> Linux does, unless you have your own UNIX box. And if you did, then of
> course you would need a second computer to do all the other stuff (and I
> mention again that I am not writing about games here). So you need Linux
> and another OS. That's the problem. If you dual boot, it's an easy
> problem. If you don't, it's a very expensive problem. So, I would guess,
> MS gets business from most Linux users, adding to its already shameful
> profits, and Linux goes along not having drivers that it needs.

Not true. To make full use of your computer you can do it with one OS:
linux or M$ or whatever suits what you need.

There are things that you have to do differently in linux vs. M$ but, they
can be done. If you want M$ Outhouse in linux, it's coming by another
name. But, I'd suggest that if that's you need, you want appearances, not
usefulness. What is Outhouse? It's a mailer (sylpheed, NS-mail, kmail,
xfmail, pine, elm, balsa, etc are also), a calendar (kde and gnome both
have them), a contact manager (kde has one, I know i've seen others), etc.
You have those available. Just not in a single product (until evolution or
tradeclient gives it to us).

Need a virus checker? Pay for it (same as in 'Doze) and you have it. Use a
scanner? I've used a microtek scsi and sometimes a microtek usb scanner.
Need a tablet? Had one hooked up for awhile. Printers, network cards, TV
cards, accelerators, trackballs, cd-writers, and more are all available.

BUT: you have to make sure the product you are going to use is going to
work before the purchase is made.

Winmodems won't (in general - I know there are exceptions) do it because
they're for 'Doze. Some printers won't. Some network cards won't. USB is
iffy no matter what OS one uses. Many others are the same way.

Buying cheap for 'Doze, then complaining that it doesn't work in linux
isn't a viable excuse (many use it wrongly). If you're serious about
cutting the umbilical from the host that's killing you, cut it and get
hardware that works. It isn't anyone's fault that you were sucked into the
phony free computer (after rebates and signing your life away to
Compuserve or M$) hype that is still being tossed around. You get what you
pay for. If the computer is cheap, guess what: it's cheaply made!
Complaining endlessly that nobody has spent their entire waking life
writing a driver to make you happy won't solve the problem. Write to the
manufacturer, tell them you wish they make a driver or open up the source,
then throw the reply and the product away when they nicely refuse and get
something else. Or, sit down and start writing a driver yourself. Writing
to the manufacturer and telling them you tossed their product and won't
touch another won't hurt anything. It won't work miracles either.

Just because a compatability list says that something works is not the
definitive answer to hardware. Asking people who own one is the best
choice. Some (printers come to mind) are listed as working 'somewhat' or
'partially' whatever that means. Some of these are repoted by someone
(anybody can do it) that they worked. Without listening to other real
owners, one can be stuck with a bad product. I can say anything I want on
a list about brother printers. But, I never owned one. I can report
several as working and it gets printed in a list. But, I'd be fabricating
my results and spreading it to the world. (I'm not claiming this is or has
happened, but M$ would stand to gain on such an effort because that would
tend to discredit listings of working hardware.)

> That's why I posted as I did. I'm sorry it upset some of you -- I
> worried about that and hoped it wouldn't happen, but of course it did.
> 
> I have to toss out at least one more stupid thing. Would it be practical
> and reasonable for Linux users to pay a small fee to support driver
> development by hardware suppliers? I'm not wondering about how that
> could be done, just whether the idea makes sense.

This type of thing always gets lots of support until money starts moving.
Then support drops off algebraically. I personally don't see a problem
with the idea. The implementation, however, might be another thing. In the
end, manufacturers would need to be dropped from this
pay-before-you-receive system because they'd be building a core of users
on linux over time that should lead to no more need for subsidy. If, after
a given period of time, they still require support because their user base
is too small, they should probably be dropped anyway rather than
subsidizing them forever for a limited effect (kinda reminds me of welfare
and public television [flame retardent suit on]).

-- 
Excuse my english. I went to US public school.