store for Linux-compatible hardware??

Hi folks,

I've been going through the list of compatible
wireless PCMCIA cards at

http://www.linux-wlan.org/docs/wlan_adapters.html.gz

and I have to say that I am very disappointed
that many of the cards listed are not available.
The ones that are, are expensive.

I am just curious but has any clever person
realized what a pain it is to shop for hours
looking for stock on Linux-compatible hardware
and simply opened up a Linux centric store?

I hope so.

Thanks for any info-
V

Vespasian wrote:
> Hi folks,
>
> I've been going through the list of compatible
> wireless PCMCIA cards at
>
> http://www.linux-wlan.org/docs/wlan_adapters.html.gz
>
> and I have to say that I am very disappointed
> that many of the cards listed are not available.
> The ones that are, are expensive.
>
> I am just curious but has any clever person
> realized what a pain it is to shop for hours
> looking for stock on Linux-compatible hardware
> and simply opened up a Linux centric store?
>

As far as I know, the linuxhardware.com domain is still
for sale. Now's the opportunity to establish that
business you always wanted!

Question: Given the extremely low margins and high
competition on hardware... would you willing to pay
.... let's say a 25% premium? While you might
answer yes, in today's world where people won't
support free software development with even a penny,
my guess is the majority of folks will answer: No.

Chris Cox wrote:

> Question: Given the extremely low margins and high
> competition on hardware... would you willing to pay
> ... let's say a 25% premium? While you might
> answer yes, in today's world where people won't
> support free software development with even a penny,
> my guess is the majority of folks will answer: No.

All I know is, I've spent probably 8 hours and
much gasoline already trying to find a wireless
card that is compatible with Linux. I've had
to take several back to the store.

I had a similar experience when I was shopping
for an ethernet card, taking a few back.

What a waste.

Some person should buy up a bunch of known-working
PCMCIA cards and sell them 1 by 1, charging
a $10 markup and maybe $10 to ship.

Vespasian wrote:
> Chris Cox wrote:
>
>> Question: Given the extremely low margins and high
>> competition on hardware... would you willing to pay
>> ... let's say a 25% premium? While you might
>> answer yes, in today's world where people won't
>> support free software development with even a penny,
>> my guess is the majority of folks will answer: No.
>
> All I know is, I've spent probably 8 hours and
> much gasoline already trying to find a wireless
> card that is compatible with Linux. I've had
> to take several back to the store.
>
> I had a similar experience when I was shopping
> for an ethernet card, taking a few back.
>
> What a waste.
>
> Some person should buy up a bunch of known-working
> PCMCIA cards and sell them 1 by 1, charging
> a $10 markup and maybe $10 to ship.

If all this is about is wireless... there's
a plethora of information about what cards
work and what don't. With ndiswrapper, a large
amount work now.

While your plan sounds nice, you'd be one
of probably 5 total customers.. .not enough
to sustain a business.

I hear your frustration... it's just not a simple
as you make it sound.

Chris Cox <(E-Mail Removed)> writes:

> If all this is about is wireless... there's a plethora of
> information about what cards work and what don't.

The problem is that the vendors keep replacing the chips every two
weeks, without this reflecting in the name of the card, and sometimes
not even in the PCI ID. I heard of a card that was said to work if
the box was labeled "made in taiwan", but wouldn't work if an
otherwise identical box was "made in china". Because of this, the
compatibility lists are constantly outdated.

> With ndiswrapper, a large amount work now.

Ndiswrapper is part of the problem. Since it works fairly well, it
reduces the incentive for someone to write native drivers. Remember
that ndiswrapper is useless on Macs, and possibly also amd64 (not
sure).

I was fortunate enough to get my hands on a 3com prism54 based card,
while the prism54 chips still worked.

--
Måns Rullgård
(E-Mail Removed)

In comp.os.linux.networking M?ns Rullg?rd <(E-Mail Removed)>:
> Chris Cox <(E-Mail Removed)> writes:

>> If all this is about is wireless... there's a plethora of
>> information about what cards work and what don't.

> The problem is that the vendors keep replacing the chips every two
> weeks, without this reflecting in the name of the card, and sometimes
> not even in the PCI ID. I heard of a card that was said to work if
> the box was labeled "made in taiwan", but wouldn't work if an
> otherwise identical box was "made in china". Because of this, the
> compatibility lists are constantly outdated.

>> With ndiswrapper, a large amount work now.

> Ndiswrapper is part of the problem. Since it works fairly well, it
> reduces the incentive for someone to write native drivers. Remember
> that ndiswrapper is useless on Macs, and possibly also amd64 (not
> sure).

Full ack, looked around not that long ago for some working PCI
wireless card which wouldn't cost a fortune, since it's only for
the fun of it. Sure the driver had to be GPL. Would be nice if it
came with a kernel.org tarball, perhaps in one of the next
kernel?

modinfo -l rt2500
GPL

Found cards with ralink chipset most usable (GPL driver) + nice
config (GUI) tool.

Network controller: RaLink Ralink RT2500 802.11 Cardbus Reference Card

--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo (E-Mail Removed) | perl -pe 'y/a-z/n-za-m/'
#bofh excuse 118: the router thinks its a printer.

Michael Heiming <michael+(E-Mail Removed)> writes:

> Full ack, looked around not that long ago for some working PCI
> wireless card which wouldn't cost a fortune, since it's only for
> the fun of it. Sure the driver had to be GPL. Would be nice if it
> came with a kernel.org tarball, perhaps in one of the next
> kernel?
>
> modinfo -l rt2500
> GPL
>
> Found cards with ralink chipset most usable (GPL driver) + nice
> config (GUI) tool.

I want it to work with the standard wireless-tools. I can't see why
many vendors produce drivers with their own screwed interfaces.

> Network controller: RaLink Ralink RT2500 802.11 Cardbus Reference Card

I looked at that driver a few months ago, and there was something
about the looks of the code I didn't quite like. I can't really say
what it was, it just looked messy somehow, just like almost every
other vendor-supplied Linux driver I've looked at.

--
Måns Rullgård
(E-Mail Removed)

Måns Rullgård wrote:
> Chris Cox <(E-Mail Removed)> writes:
>
>>If all this is about is wireless... there's a plethora of
>>information about what cards work and what don't.
>
> The problem is that the vendors keep replacing the chips every two
> weeks, without this reflecting in the name of the card, and sometimes
> not even in the PCI ID. I heard of a card that was said to work if
> the box was labeled "made in taiwan", but wouldn't work if an
> otherwise identical box was "made in china". Because of this, the
> compatibility lists are constantly outdated.
>
>>With ndiswrapper, a large amount work now.
>
> Ndiswrapper is part of the problem. Since it works fairly well, it
> reduces the incentive for someone to write native drivers. Remember
> that ndiswrapper is useless on Macs, and possibly also amd64 (not
> sure).

In a way that's a good point. BUT... experience seems to indicate
that the vendors aren't going to support Linux and it may take 20 years
or more to get support for many of those cards (which will be obsolete
in less than a year... a lot was said there... do you see the picture?

Hardware changes way to fast. Unless the vendors will open up or
drastically help in the development of free drivers, things like
ndiswrapper actually become the BETTER solution... you see?

I know.. .certainly NOT in the spirit of free software, but I'm
strictly talking about solutions. I suppose the RIGHT thing to do
is to form a free hardware company that PRODUCES hardware that
is free and open from the start. If anyone has a sustainable
business plan for that, let us know.


>
> I was fortunate enough to get my hands on a 3com prism54 based card,
> while the prism54 chips still worked.
>

Chris Cox <(E-Mail Removed)> writes:

> Måns Rullgård wrote:
>> Chris Cox <(E-Mail Removed)> writes:
>>
>>>If all this is about is wireless... there's a plethora of
>>>information about what cards work and what don't.
>> The problem is that the vendors keep replacing the chips every two
>> weeks, without this reflecting in the name of the card, and sometimes
>> not even in the PCI ID. I heard of a card that was said to work if
>> the box was labeled "made in taiwan", but wouldn't work if an
>> otherwise identical box was "made in china". Because of this, the
>> compatibility lists are constantly outdated.
>>
>>>With ndiswrapper, a large amount work now.
>> Ndiswrapper is part of the problem. Since it works fairly well, it
>> reduces the incentive for someone to write native drivers. Remember
>> that ndiswrapper is useless on Macs, and possibly also amd64 (not
>> sure).
>
> In a way that's a good point. BUT... experience seems to indicate
> that the vendors aren't going to support Linux and it may take 20 years
> or more to get support for many of those cards (which will be obsolete
> in less than a year... a lot was said there... do you see the picture?

The existence of ndiswrapper makes it more difficult to put pressure
on vendors to support Linux. They'll just point to ndiswrapper and
tell you it works, happily ignoring the fact that you're not using a
PC.

> Hardware changes way to fast. Unless the vendors will open up or
> drastically help in the development of free drivers, things like
> ndiswrapper actually become the BETTER solution... you see?
>
> I know.. .certainly NOT in the spirit of free software, but I'm
> strictly talking about solutions.

Me too. I want a solution that works all the 22 architectures Linux
runs on, or at least the major ones, i.e. x86, amd64, ppc(64),
sparc(64), alpha.

It's also a matter of confidence. How do we know what those windows
drivers are really doing? Is there any assurance they won't go
stomping all over places they don't belong at all? Even with a binary
driver specifically for Linux, we can at least be fairly certain that
the vendor is trying to play by the rules. Not so when the driver is
made for another OS entirely, and only being used through a hack.

--
Måns Rullgård
(E-Mail Removed)

Måns Rullgård wrote:
....
> The existence of ndiswrapper makes it more difficult to put pressure
> on vendors to support Linux. They'll just point to ndiswrapper and
> tell you it works, happily ignoring the fact that you're not using a
> PC.

I doubt it seriously. I mean... we're talking about less than
1% business for that vendor. The additional cost to him (well
at least perceived cost) isn't it worth it. Now... get that
number to 10% and you'll gain the attention of people in highly
competitive markets. Possible? Who knows. Linux is certainly
making gains.... other alternatives help this as well.

....on ndiswrapper
>
> It's also a matter of confidence. How do we know what those windows
> drivers are really doing? Is there any assurance they won't go
> stomping all over places they don't belong at all? Even with a binary
> driver specifically for Linux, we can at least be fairly certain that
> the vendor is trying to play by the rules. Not so when the driver is
> made for another OS entirely, and only being used through a hack.
>

Possibly true. Of course, most people using Linux aren't compiling
everything from scratch... so the knife cuts both ways. We trust it
more because we BELIEVE that it was compiled from source, and hopefully
the source delivered with our favorite distribution.

I hope we DO see more native drivers for Linux. But it would really
be nice for vendor's to see extremely tangible benefits.

Let me give you an example. Let's say a vendor comes out with a
PC that is guaranteed to work with Linux (that's a hard thing btw
since Linux is itself a moving target). Is everyone in the Linux
community going to buy that machine? Of course not. How about
50%? Of course not. How about 10%? Highly unlikely. How about
1%? Close, but probably not.

I think we need more radical change. HW vendors that start
out open and stay open. People who are not embarrassed by
their own code (e.g. Nvidia and ATI)... people who appreciate
making their products better.