Forced obsolescence in Linux

[Victor Odhner]

Joshua Zeidner wrote:
 > First off, Microsoft is making that claim.

That doesn't make it false.  I made an off-handed comment to the
Slashdot post that was contradicting Microsoft's claims.  That
post pointed to a very brief article that just said, "I found some
distros that work OK on old stuff."  But my complaint was based
on my own experience, having gone through a number of modern
distros that would not talk to this sound card, and said annoying
things such as "No soundcards detected".  Gimme a break!  Go
to a little trouble, you can see that *something* is there, even if
you can't make a noise with it.

Oh and by the way, XP can run this old 16-bit sound blaster.
So there you go.  (Never mind that I can't resurrect his XP setup
because I don't have the original disk or the key.  Reminded me
why Stallman rebelled in the first place!  But I digress.... )

Siri Amrit Kaur wrote:
 > Have you tried Vector or one of the distros that's optimized for
 > older hardware?  I think Vector and Slackware still use an older
 > 2.4.* kernel by default. Maybe it would have drivers . . .

Similarly, Dennis Kibbe wrote:
 > I know PLUGers that have installed Slackware on old 486
 > ThinkPads. ... Many distros (like Slackware, for instance)
 > have versions going back to release 1.0 still on the servers.

I don't want a different kernel, I want to add a driver.
Like I said, I am not interested in distro shopping.  This distro
does most of what I want done.  The current kernel is set up to
have drivers added, and the driver is the only change I want to
make.  Why regress to an older kernel, with bugs that have been
fixed?  Why can't I just plug in a binary driver, as with Windows?

Suppose I tried an older distro that worked fine with the sound
card, but complained that it would not read my USB thumb drive.
I'm afraid I would have heard the same thing:  Try another distro,
or upgrade the kernel to get that support.  My complaint is that
upgrading a kernel *loses* the support for the sound card.
Ya can't win fer losing.

... Which takes us to Kevin Brown, who wrote:
 > at some point the devs do have to stop trying to keep lesser used
 > drivers up to date with a newer way of doing things.

Are you saying the older drivers don't plug in the same way that the
newer ones do?  Why would they invalidate an API that was working?
Note, Windows XP did a lot of that, for security reasons; but those
were old drivers that required dangerous privileges.  I would have
thought Linux drivers would be safer from the start.

My lack of this knowledge is probably the reason for my annoyance.
I'm a developer myself, and I understand that it's hard to make an
omelet without breaking some eggs.  But when we use "insmod", that
takes the specified driver and links it into the kernel, presumably
using a somewhat generalized API.  Right?  Does each driver that
the kernel's willing to accept require some additional ugliness in the
kernel, creating a cost that has to be limited by banning some old
drivers?

I'm also assuming that the drivers consist of rather primitive and
self-contained code, so that they do not need to interface with the
system's standard libraries.  Even if they did, hopefully those APIs
would remain valid.

I'm willing to learn . . .