[ILUG] Help with nVidia drivers install FC2

[Paul Jakma]

I'll try keep this short.. i suspect it'll difficult.

On Wed, 25 Aug 2004, Bryan O'Donoghue wrote:

> On the one hand you say that Open source exponents can't live in 
> the real world without compromising sometimes, yet equating using 
> the Nvidia driver (for example) to that stated belief on your part, 
> doesn't seem to equate. Perhaps that's overharsh, but that seems 
> like a logical disparity.

The *only* thing I've pulled you up on was your claim that you'd 
looked for open-source supported hardware before buying the hardware, 
but couldnt find it.

Your reply to that was that you werent able to find any in Tralee. I 
personally find your fear of online-retailers to be unjustified. 
Further, I harbour a slight suspicion you didnt bother at *all* to 
try find, eg, a radeon S-Video card, but that's probably just cause 
i'm a suspicious, mean-spirited bastard. To quote:

"That said, if I could find a card and an open source driver for the 
card which would do the same thing under Linux, I'd vote with my feet 
... "

When I doubted above, you clarified it as, to paraphrase:

"well, i meant, if there'd been some kind of PCI card with S-Video 
out supported by Linux in a shop in Tralee on the day(s) I went into 
town there, I'd have bought it. All I wanted was a card with decent 
Linux support which I know NVidia has."

In my book, that isnt much of an effort to find open-source supported 
cards, arguments about online shopping not withstanding, where we'll 
just have to agree to disagree.

> Not true. While *you* might have a good experience buying stuff 
> from the internet,

Ra ra. There's really no point getting into an argument about online 
shopping on ILUG.

> the buying period), was the Nvidia. Irrespective of the OS, to be 
> used, the Nvidia was my only choice.

Right, in that case fair enough.

[snip lots of "hazards of buying from the internet" stuff]

> So would it be fair to say you're proposition is that running 
> Microsoft Word on Linux or even Microsoft packaging and 
> distrubuting Microsoft Word-Linux is fine, doesn't violate the 
> principals of "Openess and control" you've stipulated thus far, 
> while the Nvidia driver does?
>
> To me, that is logically spurious.

>From a technical POV, Microsoft Word for Linux would not run with 
full privileges, as, eg the NVidia drivers (both kernel and 
user-space) do.

And yes, to me there is a distinction between the underlying OS 
infrastructure and applications, the lower something lies in the 
tower of code that makes up your PC runtime, the more important it 
is. However, this is a subjective matter.

> Either a system is Open or it's not, after that all that is 
> happening is an argument over a shade of grey.

Yes, it's different shades of grey. The world has a tendency to be 
like that.

> in your argument. I don't think that makes me a 'heretic' but, I do 
> think that the thought process is logical. Perhaps you don't and if 
> you don't the please do point out the flaw.

You're right, VESA would be proprietary code bits, no different to 
Xserver XAA binary driver. However, I dont use a VESA framebuffer, 
nor do most people.

If you want to pull lots of examples of proprietary low-level, or 
otherwise, code out, go right ahead - I'm not going to bother 
replying on any subjective basis, cause yes, there's tonnes of it. 
That's a battle for another day. We do both seem to agree that it 
would be nice for these to be open too, but they're not, so let's 
live with it.

However, we're discussing one specific area: graphics cards. For 
which the battle was *almost* won, for a short while most vendors 
*did* provide docs.

Dont try to mix different classes of hardware/software. How does the 
fact there is 0 practical choice for open PC BIOS have an impact on 
deciding which graphics card to buy, where there *are* practical 
alternatives to the binary-only brands.

Stop obfuscating the argument. I cant buy a PC with an open BIOS, you 
know it, I know it. We can buy graphics cards with open drivers 
though (if only they'd sell them in Tralee!).

And Yast? Who cares? There are dozens of ways of installing Linux, 
all open. I could install SuSe Linux with cpio for ferks sake. If 
there's plenty of choice, then it does not matter if some of the 
choices are proprietary, that's fine. I dont care if others use 
proprietary, as long as it doesnt affect my preference for open.

> I accept that it *could* be. I accept that it is _not_ trusted, but, I don't 
> accept it has this flaw until you prove it!

I dont see what you're trying to argue here. I didnt claim the AP was 
trojaned. I just used its binary RTOS + Linux kernel model as an 
example of how Linux on PCs could go if we all just give up 
completely on hardware with open specs and/or driver and accept 
binary-only drivers for hardware in general, encouraging more and 
vendors to start refusing to provide docs, even for hardware which 
now is open.

And while Intersil have little interest in putting Trojans into their 
RTOS for their reference design, they just want to sell ISL3893 chips 
;), if such a model propogated to PC computing in general, there 
would be a lot more companies and people that'd we'd all have to 
simply trust.

I said nothing about things being trojaned now, I merely speculated 
on the future and contemplated the possibilities. Argument by 
extrapolation, to a future where everything is binary-only.

> Whatever you choose to call them, the magic code from Intel could 
> be Trojaned for Echelon... I can't prove that though.

I dont know about the IXP4xx, but if they're anything like the 
IXP2x00, then the microengines are fully documented and programmable, 
so if you are paranoid then you can simply write your own microengine 
code. You have the choice.

> Not logically true and you know it. Lets reduce the arugment to absurdity.

I suspect that's already the case..

> then you said
>
> Y_Driver  = B_ClosedSource
>
> Print_Accept(Y_Driver) -- would print "Logicall spurious" , which is false

I dont know how you derive that. Above is not equivalent to what i 
wrote. Forget formal logical, read this again:

>> Just because CPU microcode, or CDRW drive firmware or .... is closed 
>> source, is not a good reason to accept that your Linux kernel driver and 
>> XFree drivers for your graphics card should be closed.

Anyway, remainder of this post is pulling you on things you said 
which i suspect strongly are false/misleading.

> You are dependant on your BIOS to adequately configure your system 
> for you, so you are 'more' dependant on that then you are on, a 
> video driver.

Yes, I'm dependent on the BIOS to twiddle the appropriate northbridge 
bits, read the first 512bytes of disk 0 into RAM and jump to it (or 
jump to option ROM). After that it is out of the picture. Big deal.

Just about all of the rest of the stuff Linux can do, including load 
microcode on P6 and probably K7 and K8 soon.

> Nope, I may not view/modify the magic code for the IXP425BD, 
> without selling my soul to Intel.

I'm not 100% sure, but I suspect that's bull. They mightnt give you 
their microengine code, they will give you the microengine docs.

> How does a million IXP425BD chips with magic Intel closed code ... 
> not damage Open source? Or how does it damage it less?
>
> Why not shout the house down about that?

yada yada.. Cause it's *OPEN*.

Again, i'm almost 100% sure the microengine docs are open. write your 
own. the Xscale programming docs most definitely *are*. I have an 
Xscale PXA250 right here beside me runnning Linux, I know the IXP2x00 
run linux, as does the IXP4xx.

Again, Intel are usually good about docs.

NB: I used to have Brian.ODonoghue at kbs\.ie kill filed. Not sure if 
that was you. I think I'll have to kill file again, if only to spare 
the list, because i'm a sucker for trolls..

regards,
-- 
Paul Jakma	paul at clubi.ie	paul at jakma.org	Key ID: 64A2FF6A
Fortune:
"And remember: Evil will always prevail, because Good is dumb."
-- Spaceballs