Drivers for Linux: Great article

[gowator]

kjnowing how to compile is as simple as pressing a button if you like but is the problem the fact its compiling or the fasct the user has to drop to a CLI?

I have used MacOS-X and it does this because like *bsd obviously its a microkernel. Linux is not a microkernel and the development is not controlled enough to do this. Mac do it very well because they control the HW...*bsd do it through tighter development control.

Yet *bsd has far less users than *linux and is far behind it in opther areas of usability because of this. (Its also ahead in other areas)

The point is it is the kernel dev model that is linux....
*bsd is simple to install... you pay an expert... after that you only need to add a few drivers now and then...

this to me is like complaining that knoopix does hardware detection every boot... of course its a live CD and that is the nature of knoppix.

Linux is really all about the kernel....
I know that sounds corny but its really true!

Its often said that *bsd is a programmers/developers OS, designed as a programmer would wheras linux is a hackers OS...
Its pretty true and the hackers roots of linux still show .. but at the same time are what give linux its dynamics!

[user unknown]

Having a look into /usr/src/linux/arch, I find 21 platforms running linux.

alpha arm26 h8300 ia64 m68knommu parisc ppc64 sh sparc um x86_64
arm cris i386 m68k mips ppc s390 sh64 sparc64 v850

Today, you need one source for all platforms.

You might distribute 21 binaries, which increases the size of the distribution enormously.
But why not just discriminate 20 platforms, and ship the i386-drivers only?
It's the most used one, and other users may still get the sources somewhere!

OK, nobody may set a --debug-enabled=TRUE-flag on compiling, or optimize for his i686, or ...
All these people can get the sources - if there are any.

If we establish a culture of binaries, and it's doing well, vendors might keep the sources closed.
Get Distri-xy-9.1! There you find a working driver! might and will be the answer of a hardware-manufactor.
While you need Distri-yz-8.7 for some other hardware.

To avoid dependencies with a specific libc-x.y - version, every module could be linked static.
Buy some bigger harddrive and more RAM!
Problem solved!

In the end, we will find two different linux-flavors: The experts using source, and the newbies using binaries.
But there will be no bridge to help the newbies. The advices from experts will not work for the newbies.

And of course you need to buy a new distribution for every new hardware, and reinstall the whole system for every major upgrade.
Welcome to the world of windows!

[j.drake]

This is interesting.

http://news.com.com/Linux+standard+g...l?tag=nefd.top

Drivers, hell! Looks like there's a movement to standardize Linux!

jd

[user unknown]

I don't see binaries covered by the article, and even drivers aren't mentioned.

don't visit my homepage:http://home/arcor.de/hirnstrom/

[Flash00]

This is interesting.

http://news.com.com/Linux+standard+g...l?tag=nefd.top

Drivers, hell! Looks like there's a movement to standardize Linux!

jd

Yeah! Standards are great; there are so many to choose from!

Seriously, I think a movement to standardize Linux is probably a good thing, but I think it ought to come from ISO or IEEE, or even NIST, rather than an industry group. I don't have any problem with companies making money off Linux or with them contributing to the Linux community, but they are in business to make money, so they may want to take Linux in a direction the rest of us don't like. How to make sure that the camel stays outside the tent?

[j.drake]

Sometimes one has to make sacrifices for the greater good. Linus could do this himself if he chose to, or at least give his official blessing. Of course, he could also stick to his principles and refuse. The cost of refusal is further fragmentation, which hurts linux and helps Microsoft. But, arguably, standardization could stifle innovation and alienate some of the most enthusiastic supporters.

Right now, we have one fragmented standard. If something isn't done, we will have two standards - one for the commercial interests and one fragmented standard for the hobbyist interests. Not necessarily a bad thing - Red Hat effectively did that when they spun off Fedora. Doubtless, most hobbyist distros would incorporate the commercial standards anyway, and deviate from them as necessary to provide specific enhancements or needs. Maybe it's a good thing, and the distros outside the commercial camp will be like a big R&D department providing much of the innovation.

jd

[gowator]

We already have sacrifices for the greater good but from what you say about linus I feel you prefer the BSD development model.

This is very structured and centrally controlled and its also already a microkernel.

You make it sound simple to rewrite linux as a microkernel but I get the feeling you have never written a driver or modified the kernel source yourself (just like the author of the article).

I have tried or at least contributed ...

What you seem to be forgetting is you seem to want the same inary drivers which run on everything from sparc to G5 ...not to mentionb handheld devices etc.

WinCe is a completely seperate OS to 98 or 2000 or XP...
In linux it doesnt matter if its a palm/G5 or SGI supercomputer ...
if you compile from source it works... (+/- a bit of fiddling)

[j.drake]

I get the feeling you have never written a driver or modified the kernel source yourself (just like the author of the article).

You are right. I'm just a Windows user who has been using linux for a little over a year, but I don't think that exactly disqualifies me from offering an observation on the direction that linux needs to take in order to be a viable desktop replacement for Windows, at both the home and enterprise level.

With all due respect, it appears to me that the linux community has been altogether too dismissive of the ideas and needs of the millions of others like me who have looked at the viability of linux as a desktop alternative to Windows, and concluded that it isn't ready yet. The attitude of "you have'nt ever coded linux before, therefore your input is irrelevant and your opinions about the direction linux should take don't matter" is positively dripping form many of the comments in this thread, and altogether too pervasive in the linux community. Wise up. The future direction of linux is not in the hands of the linux elitists.

jd

[CrashedAgain]

Just to add my two_whatever_currency_you_want.
j. darke & probono, you are right on.

Ease of use should not be a privilege of paying users.

[j.drake]

The wifi problem is a special problem for opensource and is due to FCC regulations .. so dont blame linux blame your government regulators.

Most people choosing an OS don't care who's to blame - they only care what works. Blaming doesn't get anyone anywhere.

Besides, I don't buy it. There are drivers all over the place. In my case, it's a prism2 driver, which is one of the more common ones out there for linux. What's more, the driver is sitting right there on the Knoppix disk, but Knoppix is looking for it in the wrong place and can't find it. I've posted this bug for three revisions straight. You may not think it's a big deal, because you can edit, recompile, remaster, what have you. But people who need wi-fi won't bother unless they're hacking for a hobby.

Excuses are for losers, and the linux community needs to decide whether or not it's satisfied with pointing fingers and making excuses, or whether it wants the OS to reach its potential.