Is it really that cost Ineffective to produce easy to install cd-rom hardware drivers for Linux ** STILL ** ?

Suppose that there were 5 million desktop Linux users in the world.and suppose further that the major hardware companies produced easy installation cd-rom drivers for a small range of ordinary Printers, Winmodems, Soundcards, Scanners etc.etc.etc. Suppose further that it became "well known" amongst desktop Linux users that a range of Epson/HP printers/Intel ambient winmodems/Creative soundcards/Canon scanners etc etc etc came with easy to install cd-rom drivers for Linux. Wouldn’t it be the case that a significant percentage of desktop Linux users would flock to the Hardware manufacturers who produced easy to install cd-rom drivers for many of their peripherals.(I would!) Suppose that only 1 million of the 5 million desktop Linux users bought the relevant hardware from the manufacturers who produce easy to install cd-rom drivers that support a certain kernel. Wouldn’t that be cost effective in terms of the balance between the cost of producing a driver against the revnue brought in by the people buying the niche "well known Linux" hardware. How much does it cost to create a driver for a printer, winmodem, soundcard, scanner?

Suppose Canon sold 55,000 extra printers at $149 a printer due to the added Linux driver - would that be cost effective in relation to Canon producing the driver in the first place? How many desktop Linux users are there WORLDWIDE? What percentage of them would flock to the niche "Linux supported hardware" and buy it? (I would!)

I suspect that there is a mentality amongst a significant percentage of ordinary desktop Linux users whereby they *would* flock and *buy* Linux supported harwdare containing easy to install cd-rom drivers for their peripherals.

I also suspect that its cost effective already for a limited range of hardware. Of course there’s obscure drivers on websites and obscure things you can do behind the scenes to get your peripherals working (I'm not trying to dis this), but I think its fair to say that for the ** Majority ** of ordinary desktop Linux users they would prefer an easy to install cd-rom driver.

So to summarise – I suspect that the numbers are already there (and certainly will be if we take it 5-10 years on) that the mentality is there to flock and buy and that its cost effective already for a limited range of hardware. – so why aren’t we seeing more drivers on more cd-roms. Is it really *still* that cost ineffective? or are there any other reasons which cloud matters?

you are right the only reason that i have a nvidia card on my box is that they have good support for linux. also when ý bought my fýrst usb disk i choosed between 3 disk i did not know all were goýing to work but i got the one which said it was compatible with linux.

It depends... some kit just has too small a market to even warrant driver revs of buggy windows drivers let alone linux drivers. I see a lot more Linux drivers these days though: Matrox, Nvidia, and ATI...

Nearly every SCSI card and ATA raid controller has vendor produced drivers.

Cisco home spins drivers by the bushel, but hey... you're paying for them.

Recently I've found two wireless chipset makers, Realtek and ADMtek have produced some surprising in-house drivers.

The big deal is that its not just a matter of providing drivers... that won't work for most people, who are going to need the source code, because a binary pre-compiled module for Redhat 8.0 and its 2.4.18 kernel and 3.1.x based gcc is useless on my 2.4.21-xfs kernel with gcc 3.2.2. Then they have to make certain that source compiles right over a range of kernels and compilers and glibcs... instead of windows, which is the same thing on every machine out there... the same basic kernel and build.

The halfway point is usually where things work out very well... again, like Cisco, Via, AMD, or 3com, who share all the specs a kernel hacker wants so both win in the end, the driver gets written and maintained by someone that knows what they are doing, and the big company doesn't have to pay for it... and better for them, doesn't have to support it.

Cheers,

Finegan

Hi Finegan

>> The big deal is that its not just a matter of providing drivers... that won't work for most people, who are going to need the source code, because a binary pre-compiled module for Redhat 8.0 and its 2.4.18 kernel and 3.1.x based gcc is useless on my 2.4.21-xfs kernel with gcc 3.2.2. <<

Surely if we look at the probable uptake of desktop Linux WORLDWIDE over the next 5 - 10 years then "most people" won't need the source code - suppose at the moment a large percentage of Linux users are hardcore enthusiasts and that the distros that they are using are coming out every 6 months or so - obviously these people know how to use the source code and indeed would have to, to make a driver work as the kernels change - but isn't it more likely over the next 5 - 10 years that the larger percentage of Linux users WORLDWIDE are going to be "ordinary users" and further that the major distros that these ordinary users would be using would be released at a much slower rate - maybe every 1 year or 1.5, 2 years - if these things hold and thats the case - then surely its worth looking at the idea of a Canon printer driver that is compatitble with say Red Hat 10,s 2.x.xx kernel for the duration of say 1-2 years for which the major desktop distros of the future might span.

I don't mean to sound disrespectful to the way things have been done in the past by experienced users - I am just thinking in terms of the likely growing numbers of ordinary users over the next 5 - 10 years that IMO would flock to and buy a specific Canon printer or a specific modem or soundcard etc etc if they knew in advance that it had an easy to install cd-rom driver for the kernel in the distribution that they are using.

Skyline