[COLUG] Re: Video Cards

[William Yang]

Rob Funk wrote:
> And not just practicality for those who actually have nVidia cards 
> (whether you care about binary drivers or not), but also those of us who 
> prefer to stick with the advantages of open drivers.

All things being equal, I'd probably select an open driver over a
closed one.  Unfortunately, not all things were, or are, equal.  Priorites 
(such as religious fervent) are in play.

> That's where the kernel developers' statement, and the detail in the "Tale 
> of Three Drivers" essay, come in.  The binary drivers *do* have bugs.  

Non-persuasive argument.  You can tell me the code's the worst
code since whenever, and I would only care if the bugs affect the needs of
the systems I'm using those components in.  Your point?

There are plenty of oopses that don't include the ATI and nVidia closed 
drivers.  People can choose not to use these things if the bugs are too 
onerous; or they can accept the risks as acceptable (as I and others do).

> People *do* complain to Linux developers about those bugs.  The Linux 
> developers can't do anything about it.  Or, the kernel has bugs that 
> appear when people use the binary drivers, but the developers still can't 
> do anything about it without the driver source.

And?  You're acting like this kind of thing hasn't been happening since
before the first C compiler.  When you have different systems that have to 
interact, and those systems have bugs, you often get finger pointing and 
complaints going to the wrong place about components.  I don't see this as 
being a problem of open v. closed software: it's just a problem that's 
common to complex software systems.

> Further, we get incidents like Fedora being pressured to delay a release 
> because nVidia hadn't yet released a driver compatible with the version 
> of X in the Fedora release.  http://lwn.net/283291/

That link didn't work for me.

I won't say I'm privy to many details about the situation, but it's my
understanding that nVidia said they would release a new binary driver on
their own schedule, and Fedora said they were going to release with certain
software versions requiring that new binary driver prior to that date.
Fedora had a choice, and they exercised it.  nVidia had a choice and they
exercised it.  This all sounds pretty free (as in speech, specifically) to 
me.  Frankly, it sounds more like Fedora tried to pressure nVidia from what 
I've read, and that they were unsuccessful.

Unless you're telling me nVidia tried to get a restraining order or some
other form of actual, external restraint on Fedora (which I'd think I would
have heard about), this sounds like you're getting upset because some
people didn't agree with one another.

> Or analogously, the kernel developers are constantly getting pressured not 
> to break binary module compatibility as they improve things, while 
> without such huge popularity of a small number of binary drivers they'd 
> be more free to change internal interfaces as necessary.

You obviously value the independence of each component that goes into a
Linux distribution -- and of each distribution itself.

Unfortunately, I see these values as being in conflict:  if you reject the
underlying assumption is that everyone has exactly the same goals in terms
of deadlines, budget/available resource allocation, etc (in this case, a
claim that's appears likely to be demonstrably false), the conflicts become
apparent pretty quickly.  If you don't have that underlying assumption,
it's a logical consequence (and an exercise to the reader) that conflicting
goals will give rise to conflict that can readily encourage the use of
pressure, tactics, negotiation, and worse.

> Did you read any of the links...

I'm just not as easily convinced by written opinions which are not
backed by facts, even the opinions of those who write code I use.

The market forces may or may not work as described in 
https://www.linuxfoundation.org/en/Linux_Graphics_Essay.  There are other 
rational ways to respond to the high cost of operation beyond opening the
drivers.  It all depends on the valuation of the intellectual
property involved.

	-Bill
-- 
William Yang
wyang at gcfn.net