• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Wireless Internet for Linux

Or do some research and get a well supported card so you don't have to worry about stupid hacks like ndiswrapper.

RaLink cards are supported without any binary-only firmware and completely free GPL'd drivers. The drivers aren't included in the base kernel yet, but chances are you'll have an easier time getting them working than ndiswrapper.

http://ralink.rapla.net/
 
You shouldn't have to buy a wireless card based off of whether it will work with the OS. The OS should be designed to handle the hardware.

And to clarify, NDIS wrappers are not stupid.
 
Yes, you should. Just as you should do some research on the quality of the device itself before you buy it. Would you recommend that someone should just go and by any car stereo before finding out how much of a PITA it's going to be to mount the head unit? I can't imagine anyone would utter the words "You shouldn't have to buy a head unit based off of whether it will work with the car. The car should be designed to handle the hardware.".

In an ideal world things would be a lot simpler, but with companies like Broadcomm out there that put out crap hardware and won't release any specs on that crap hardware or even release Linux drivers themselves it's impossible to say "the OS should just support it"

You should be supporting companies that support Linux, only then will hardware companies realize that supporting Linux is a good idea for them. Luckily most already have, but there are some stragglers out there that just don't get it.

And I hope ndiswrapper dies a fast and nasty death. All it does is let companies know that they don't have to make their cards accessible because Linux users can simply use the Windows drivers. Why waste time developing drivers or put out a doc on how the hardware works when you can just let people emulate what you do support?

There was a recent thread on lkml about making 4K stacks the default for the kernel instead of the 8K that's used now. It seems that Windows uses a 12K kernel stack, or higher I'm not sure, so it's already hit or miss as to whether Windows drivers will work with ndiswrapper. Once the change is implemented it'll be virtually impossible for ndiswrapper to work without a major rewrite of the project. Some people had reservations about the change for other, better reasons. But most of them were of the "that's dumb, don't do that" type problem and could easily be worked around.
 
Originally posted by: calculusz
You shouldn't have to buy a wireless card based off of whether it will work with the OS. The OS should be designed to handle the hardware.

You misunderstand the nature of the problem.

Part of the reason for this is because you are likely very used to being solely dependent on the hardware manufacturers for drivers. That is not how things work in the FOSS world.

And to clarify, NDIS wrappers are not stupid.

Perhaps "stupid" is too vulgar a term; perhaps "incredibly inefficient and pointless" is better.

Have you ever tried to reverse-engineer anything? I guarantee, it gets really old, really quickly, when there isn't a good reason for it. And there is no good reason not to give us hardware specs.
 
NDISWrapper is the single most valuable package for me for the plain and simple fact that Linux wireless support sucks..... period.

<Cue zealot flames>
 
Time is the most valuable for most 😉

Seriously out of 8 major wireless NIC's I have trialled with linux (we are talking major manufacturers and chipsets - the kind the average consumer is given the choice of when walking into a store)

ZERO are supported out of the box
2 Work With NDISWrapper
6 Dont work no matter what and believe me 3 weeks of scouring mail lists and forums is not fun research especially when even the supposed Linux Guru's are unable to help.
 
Seriously out of 8 major wireless NIC's I have trialled with linux (we are talking major manufacturers and chipsets - the kind the average consumer is given the choice of when walking into a store)

That's because most of them reuse the same 2 or 3 chipsets and sadly Broadcomm is pretty popular and they're probably the worst out there. I bought my RaLink card at CompUSA, with a little reading you too can own a card that doesn't suck.

The Linux kernel developers don't care about ndiswrapper or the closed nvidia driver or the closed ati drivers, they're all used with a YMMV stamp on them. The only difference I can see is that you might actually get support from nVidia about their driver but Linksys will just laugh at you if you call them asking about using their card with ndiswrapper.
 
Sigh, come on seeruk. When my M-Audio drivers started to causes issues with my system, did I try to cop out and blame it on MS? No, I knew that my ire should be directed at the company responsible for developing garbage drivers. Hence I went out and bought a Audigy. No more panel applet crashing and taking out my sound.

I don't really see how the arguement on GNU/Linux has poor wireless support, when they validate certain hardward into the kernel. Above and beyond that if it's not supported within the kernel it, most likely, will be a PITA to get it to function by using work arounds or hacks.

It would be the same for me trying to use a piece of hardware that is not validated on Windows and complaining that Windows stinks because it doesn't properly support my hardware.
 
Still the fact remains that generally speaking wireless cards suck in Linux.... but you know what? I bet if the same effort went into hardware support in Linux as went into ... oooh ummm the 995th ragnarok clone, or yet another desktop, or implementing IPv6 even though IPv6 is supported only by enterprise level hardware..... a generic broadcom, atheros, whatever driver or two could be developed.

And you see there is problem, Linux is driven by the whim of the individual developer rather than a focused and concerted effort to get the basics right.

Once again I will re-advertise the fact that I use Linux everyday and love it as a secure web/application server - but having read another rant by another distro yesterday about how Linux is a better desktop than windows I think I need to get my hands on some the dillusional drugs the writer was taking and I can pretend I am superman for a day.
 
Originally posted by: Seeruk
Still the fact remains that generally speaking wireless cards suck in Linux.... but you know what? I bet if the same effort went into hardware support in Linux as went into ... oooh ummm the 995th ragnarok clone, or yet another desktop, or implementing IPv6 even though IPv6 is supported only by enterprise level hardware..... a generic broadcom, atheros, whatever driver or two could be developed.

And you see there is problem, Linux is driven by the whim of the individual developer rather than a focused and concerted effort to get the basics right.

Once again I will re-advertise the fact that I use Linux everyday and love it as a secure web/application server - but having read another rant by another distro yesterday about how Linux is a better desktop than windows I think I need to get my hands on some the dillusional drugs the writer was taking and I can pretend I am superman for a day.

emerge madwifi-drivers
modify /etc/conf.d/wirless (and you have to "Set up" your card in windows too)
/etc/init.d/net.ath0 start

that was tough...oh, and I don't yet have it working in windows.
 
Insert card
Boot Windows
<optional enter key/ESID>
Browse

That was hard....

And please explain how to emerge md-wifi drivers with no NIC & thus no internet 😛

and the fact that mad-wifi supports only a few chipsets
and in fact you may have to go the NDISWrapper route
in which case you also skipped a few steps such as install linux headers, correct GCC versions etc
all without an internet connection
.... but I'll let it slide 😛 😛
 
Originally posted by: Seeruk
Insert card
Boot Windows
<optional enter key/ESID>
Browse

That was hard....

And please explain how to emerge md-wifi drivers with no NIC & thus no internet 😛

and the fact that mad-wifi supports only a few chipsets
and in fact you may have to go the NDISWrapper route
in which case you also skipped a few steps such as install linux headers, correct GCC versions etc
all without an internet connection
.... but I'll let it slide 😛 😛


Same card, insert, boot to windows, new hardware found, cannot find drivers...crap, now I'm looking, but this is an Atheros branded card (not just chipset) no drivers there. Let me try these...nope, these...flakey. on and on.


Oh, and just so you know, emerge means I already have GCC and all that other stuff, cause it's gentoo, I'll let it slide.

apt-get install madwifi-drivers works fine, without all that stuff too.

I am talking about atheros based cards, as that is what I use. I have no experience with any others.

Also, if you have to compile ndis wrapper, wouldn't you need network to get it, and also the kernel headers, etc?
 
I bet if the same effort went into hardware support in Linux as went into ... oooh ummm the 995th ragnarok clone, or yet another desktop, or implementing IPv6 even though IPv6 is supported only by enterprise level hardware..... a generic broadcom, atheros, whatever driver or two could be developed.

If you think it's that easy why don't you start the project? I'm sure you'll find a decent amount of people willing to help since Broadcomm wifi chipsets are so popular.

And you see there is problem, Linux is driven by the whim of the individual developer rather than a focused and concerted effort to get the basics right.

And that's why Linux has had such little growth over the years, right? The development model Linux uses works, you can't deny it. Sure not everything gets 100% coverage, but the same is true of commercial software. A lot of Linux developers are working for nothing, so of course they're going to work on what they find interesting and most people find reverse engineering hardware quite frustrating and not very fun. Other companies (RaLink, Atheros, Intel, etc) have taken the small steps of releasing firmware and docs so that free drivers can be written, if Broadcomm can't be bothered to even do that then I don't want to use their hardware anyway.

Insert card
Boot Windows
<optional enter key/ESID>
Browse

You forgot:

find CD
install drivers
reboot
poke and hope for 5 minutes looking for where to enter the SSID
 
Originally posted by: Seeruk
Still the fact remains that generally speaking wireless cards suck in Linux.... but you know what? I bet if the same effort went into hardware support in Linux as went into ... oooh ummm the 995th ragnarok clone, or yet another desktop, or implementing IPv6 even though IPv6 is supported only by enterprise level hardware..... a generic broadcom, atheros, whatever driver or two could be developed.

And you see there is problem, Linux is driven by the whim of the individual developer rather than a focused and concerted effort to get the basics right.

Once again I will re-advertise the fact that I use Linux everyday and love it as a secure web/application server - but having read another rant by another distro yesterday about how Linux is a better desktop than windows I think I need to get my hands on some the dillusional drugs the writer was taking and I can pretend I am superman for a day.

Get your wallet out, pay for the development.
 
Just to provide a few counter examples:

2 nights ago, I did an FTP install of Debian 3.1 (sarge) with a orinoco pcmcia card, no problems
last night, I did an FTP install of Redhat Enterprise 4.0 (actually CentOS 4.0) with an orinoco pcmcia card, no problems
a year ago, I did an install of Gentoo with an orinoco mini pci card, no problems
3 years ago, I did an install of Debian unstable with a prismII based card, no problems

The only time I had any trouble was 2 years ago with the orinoco mini pci card, since the driver hadn't made it into the main kernel tree yet.



When I get a linux system with wireless, I know exactly how to configure it to set the essid, encryption key, etc. With windows, you have to find out if they are using XP's utility, or any of the 50 utilities from different manufacturers.
 
On the windows front, I've had pisspoor results with my centrino laptop. The wifi card absolutely stinks in there. I bought a orinoco (gold I believe) with the external port, coupled with netstumbler is an amazing duo. Now again I knew it was A) a poor antenna on the Intel card or B) craptastic software for controlling the card. At no time did I swing a finger at MS, because wifi isn't something MS is providing. They are providing the network stack for that wifi to ride on. Now if that was having a problem... 😉
 
Originally posted by: Seeruk
Still the fact remains that generally speaking wireless cards suck in Linux.... but you know what? I bet if the same effort went into hardware support in Linux as went into ... oooh ummm the 995th ragnarok clone, or yet another desktop, or implementing IPv6 even though IPv6 is supported only by enterprise level hardware..... a generic broadcom, atheros, whatever driver or two could be developed.

And you see there is problem, Linux is driven by the whim of the individual developer rather than a focused and concerted effort to get the basics right.

Once again I will re-advertise the fact that I use Linux everyday and love it as a secure web/application server - but having read another rant by another distro yesterday about how Linux is a better desktop than windows I think I need to get my hands on some the dillusional drugs the writer was taking and I can pretend I am superman for a day.


The whole idea of GNU and GNU/Linux is that it gives developers an oportunity to develop applications in a standard environment. Don't knock the developers that are doing the 996th variation of Ragnarok or the IPV6 guys because that is what they choose to develop. They aren't there to cater to YOUR ideals or whims. They are there to create and learn what they want to do freely without the M$ machine on their back trying to sell them a compiler and API set. Furthermore, desktop creators are X developers, or striving to be, whom add back into a pool of understanding about how X works which can go back into the X system calls overal improving X Windows on every system, even M$ Windows.

If you have beef with a company's drivers or an OS support go after the two or three responsible : OS manufacturer ( kernel devs AND distro publisher for linux's case ) and the hardware manufacturer. Don't belittle a community of programmers who have been freed by a product to do what they set out to do.
 
Originally posted by: Nothinman
Or do some research and get a well supported card so you don't have to worry about stupid hacks like ndiswrapper.

RaLink cards are supported without any binary-only firmware and completely free GPL'd drivers. The drivers aren't included in the base kernel yet, but chances are you'll have an easier time getting them working than ndiswrapper.

http://ralink.rapla.net/

Or thank the OP for giving this link to people who already have a card that isn't supported under Linux, and may one day want to try playing around with Linux without having to buy new hardware?

 
Originally posted by: KeyserSoze
Originally posted by: Nothinman
Or do some research and get a well supported card so you don't have to worry about stupid hacks like ndiswrapper.

RaLink cards are supported without any binary-only firmware and completely free GPL'd drivers. The drivers aren't included in the base kernel yet, but chances are you'll have an easier time getting them working than ndiswrapper.

http://ralink.rapla.net/

Or thank the OP for giving this link to people who already have a card that isn't supported under Linux, and may one day want to try playing around with Linux without having to buy new hardware?

Or how about instead of bending over for a bad vendor, getting them to release the proper docs so developers can create drivers.
 
Originally posted by: n0cmonkey
Originally posted by: KeyserSoze
Originally posted by: Nothinman
Or do some research and get a well supported card so you don't have to worry about stupid hacks like ndiswrapper.

RaLink cards are supported without any binary-only firmware and completely free GPL'd drivers. The drivers aren't included in the base kernel yet, but chances are you'll have an easier time getting them working than ndiswrapper.

http://ralink.rapla.net/

Or thank the OP for giving this link to people who already have a card that isn't supported under Linux, and may one day want to try playing around with Linux without having to buy new hardware?

Or how about instead of bending over for a bad vendor, getting them to release the proper docs so developers can create drivers.


You're not understanding. If I already bought a wireless card for my WINDOWS machine, and I can't return it. Then what, don't try to use Linux? Email the vendor telling him how mad I am at him?!? If anyone should be frustrated at this, it IS me. I can't use my wireless card at all. I even tried to use the NDIS wrapper linked above a while ago. So, when I boot into Linux, it's all wired surfing for me, for now. Which is one reason I don't use it as much. (Will probably pick up a card soon, still looking for friends that might have one supported, to trade with me.)

So, I don't understand what people are supposed to do if they already have a card not supported.



KS



 
Originally posted by: KeyserSoze
Originally posted by: n0cmonkey
Originally posted by: KeyserSoze
Originally posted by: Nothinman
Or do some research and get a well supported card so you don't have to worry about stupid hacks like ndiswrapper.

RaLink cards are supported without any binary-only firmware and completely free GPL'd drivers. The drivers aren't included in the base kernel yet, but chances are you'll have an easier time getting them working than ndiswrapper.

http://ralink.rapla.net/

Or thank the OP for giving this link to people who already have a card that isn't supported under Linux, and may one day want to try playing around with Linux without having to buy new hardware?

Or how about instead of bending over for a bad vendor, getting them to release the proper docs so developers can create drivers.


You're not understanding. If I already bought a wireless card for my WINDOWS machine, and I can't return it. Then what, don't try to use Linux? Email the vendor telling him how mad I am at him?!? If anyone should be frustrated at this, it IS me. I can't use my wireless card at all. I even tried to use the NDIS wrapper linked above a while ago. So, when I boot into Linux, it's all wired surfing for me, for now. Which is one reason I don't use it as much. (Will probably pick up a card soon, still looking for friends that might have one supported, to trade with me.)

So, I don't understand what people are supposed to do if they already have a card not supported.

KS

Make sure these vendors know that they've lost a customer. Make sure these vendors know you'll be going to one of their competitors, but all they have to do is open up the documentation to their hardware so open source drivers can be written.

Consumers should have some control over the hardware they purchased, right now we don't. We let vendors tell us what we can use and what we cannot use. It's BS. Using junk like NDISwrapper only lets the vendors know that they can continue to treat PAYING customers as trash.
 
Take this example. Four hardware RAID controller cards are evaluated amongst each other. The two cards, A and B have almost identical performance on every level. Card C gives about 80% of the performance metric, though it costs around half as much of Cards A and B. While Card D suffers from unforseen bottlenecks that can't be explained as it's using hardware similiar to that of cards A and B. Card D costs roughly the same as A and B.

Which would you choose for your system?

From a cost/performance analysis, Card C is the clear winner. It offers the best performance to cost ratio when compared to A and B. While card D is discarded as a valid choice due to the underperforming numbers that it puts up for this mock benchmark setting.

With like hardware, you would have to pin the lack of performance on poorly written drivers.

In the real world, think of how the Volari cards perform. Supposedly if we went strictly by the white papers they appeared to be a decent budget card. The hardware itself is hampered by extremely poor drivers.

Hardware and software developers have to write the API or drivers to get their products to function within their environments. Nobody ever handed Microsoft a piece of either product and said, "You make all the bucks, now integrated our product with your OS." They write and rewrite the neccesary code for either the software or hardware to function as the developers see fit. There is no magic on the Linux kernel team, or from the Microsoft camp that will make a brand new product work without issue.

Ever wonder why ATI actually is starting up their own Linux driver support? They see an opportunity to make money. Honestly they need to work on their OpenGL support in general, but that's another story.
 
Back
Top